Shift to Neutral at a stoplight. Rough in Drive, smoother in Neutral → engine mounts. Same in all gears → engine-side (plugs, vacuum, MAF). Rough only when cold, then smooth → spark plugs. Rough at idle, smooth at highway → vacuum leak or throttle body. Flashing check engine light → active misfire, stop driving now. The line that carries the page: the pattern of when it runs rough names the system — and the free Neutral test splits mounts from the engine in sixty seconds.
Your car is shaking, stumbling or vibrating when it shouldn't be — the RPM needle bounces at stoplights, the engine sounds and feels uneven. This is what mechanics call "running rough," and while it feels alarming, the cause is usually one of a small number of well-understood problems. Here's the fact that sorts them: the pattern of when it runs rough names the failing system. Rough only at idle but smooth on the highway is one family; rough in Drive but smooth in Neutral is another; rough only when cold is a third — and each points at a different, often cheap, fix. This page is about the idle. An engine that runs rough at speed is failing under load, and car sputtering under load is the page for the fuel pump, the secondary ignition and the converter.
That matters because rough running is one of the easiest problems to overpay for. The engine feels sick, so shops reach for a "tune-up" — plugs, coils, injectors — and the bill climbs into the hundreds before anyone has identified the cause. Two free tests split the whole field before you spend a dollar: the Neutral test separates a mount problem from an engine problem, and a parts-store scan names the cylinder or the condition. Injectors in particular are worth pricing before agreeing — fuel injector replacement cost runs from a seal to four figures on a direct-injection engine.
I built PulsCar after a $380 "full diagnostic" in Phuket in 2025 led to roughly $6,000 over six months across several shops, and the sound never went away (the story is on the About page). Rough idle is the textbook overpay: a shop replaces plugs, coils and injectors for $750 when the cause was a single torn intake boot. This guide teaches you to read the pattern first — the Neutral test and the free checks, all five idle causes with 2026 costs, and the trap that turns a $45 hose into a $750 parts bill.
First: mounts or the engine? The Neutral test
This one free test splits every rough-running cause into two families in sixty seconds. At a stoplight, engine warm, shift to Neutral and wait ten seconds. Does the roughness change?
Noticeably smoother in Neutral than Drive? → engine or transmission mounts (Cause 4). In Drive the torque converter loads the drivetrain and strains the failed mount; in Neutral that load lifts. One caveat: a misfire also eases a little in Neutral because the load drops, so confirm with the scan — mounts set no code, a misfire sets a P030X.
The same roughness in all gears? → an engine-side cause: a vacuum leak, the throttle body, a misfire, or a stuck EGR valve (Causes 1, 2, 3, 5).
Then read the codes free at a parts store: P030X names the misfiring cylinder, P0171/P0174 flags a lean condition, and no codes points at mounts or the throttle body. That single fork is the most valuable thing you can determine, and it's free.
The free checks: under $25, mostly free
The Neutral test. Smoother in Neutral means mounts; the same in all gears means engine-side. The decisive first split.
The OBD scan. Free at most parts stores — P030X is a misfire cylinder, P0171/P0174 is lean, P0401/P0402 is EGR flow, P0505-P0507 is idle control. The codes name the system.
The vacuum-leak listen test. With the engine idling, hold one end of a length of rubber hose to your ear and move the other end along the vacuum hoses, throttle-body base and intake seams — a leak gets suddenly loud. Keep hands, hair and the hose clear of the belts and fan. Soapy water doesn't work here: a vacuum leak pulls air in, so nothing bubbles. Never spray flammable cleaner around a running engine to hunt one — that's a fire risk. If listening finds nothing, a shop smoke test ($50-$150, per CarParts.com) pushes visible vapor out of every crack at once.
The coil-swap test. With a cylinder-specific code, swap that coil to an adjacent cylinder — if the misfire follows, it's the coil; if it stays, it's the plug. Free, and it prevents buying both.
The PCV rattle. Pull the PCV valve and shake it — a working one rattles, a silent one is stuck ($5-$20 to replace).
Why is my car running rough? The pattern of when it runs rough tells you the cause, which is why a couple of free tests beat guessing at parts. Rough only at idle but smooth while driving points at low-RPM causes — a vacuum leak, a dirty throttle body, or spark plugs. Rough at speed or under load is a different symptom — the sputtering guide's job. Rough only when cold then smooth points at spark plugs, since a cold engine needs a stronger spark. Rough with the engine visibly shaking is a misfiring cylinder. Two free tests sort most of it: the Neutral test — if the roughness eases in Neutral, mounts are the suspect — and a free OBD scan, where a P030X names the cylinder and P0171 flags lean. A rough idle is the one symptom a parked recording actually captures: PulsCar's free check returns an Engine Health Score and a likely cause from it — a second opinion, not a verdict.
What actually determines a rough running cause and its cost
Neutral or all gears — the master variable. Smoother in Neutral is mounts (a $200-$600 job); the same in all gears is engine-side (often a $0-$25 fix).
When it is rough. A hunting idle that smooths at speed is a vacuum leak; rough-when-warm with the AC on is the throttle body; cold-only with a beat is plugs; a stumble on the way down to idle is EGR. Rough at speed too sends you to the sputtering page instead.
Whether the light is flashing. A flashing light is an active misfire destroying the converter in real time — the one pattern that means stop now, not diagnose later.
Whether anyone else is paying. Thirty-one percent of the federal rough-idle complaints come from cars at or under 36,000 miles, and a third of the manufacturer bulletins on this symptom are software. Your in-service date and a VIN-level bulletin search can move the whole bill off your card.
What causes a rough idle? The most common causes, in order, start with a vacuum leak: unmetered air disrupts the air-fuel ratio at idle where even a small leak is a large share of airflow, and the tells are a hunting idle RPM and roughness that smooths at highway speed. Next is a dirty throttle body, where carbon restricts airflow at the barely-open idle position — rough idle when warm, stalling at stoplights, worse with the AC on. Third is worn spark plugs, whose misfire creates a beat-pattern shake worst when cold. Fourth is engine mounts, which fail to isolate the engine's vibration — worse in Drive than Neutral. Fifth is a stuck EGR valve, a stumble on the way down to idle. The first move is the Neutral test, then a free OBD scan. A PulsCar recording of the idle can hear a misfire's beat and a hunting idle; it can't feel a mount, which is what the Neutral test is for.
The rough running price ladder: every cause, cheapest first
Read the ladder as the argument for the free tests. The top rungs — a scan, a $15 throttle-body clean, a $10-$40 hose — fix a real share of rough idle, which is why they come first. The bottom rung is almost never a cause of rough running; the converter is the victim of an ignored misfire, so a flashing light left too long is how an $80 set of plugs turns into a $900-plus part.
The 5 idle causes, ranked by pattern
1. Vacuum leak — including the PCV valve and its hose — $15 to $400
A vacuum leak is the single most common cause of rough idle. At idle the engine needs very little air, so even a small leak is a big share of total airflow and creates a lean mixture — at highway speed total airflow is much higher, the leak becomes insignificant, and the roughness disappears. The hunting-RPM tell: an idle needle rising and falling on its own (800, 1,200, 800, 1,200) is the clearest signature, as the computer adds fuel for the extra air, overcorrects, and oscillates. The listen test (above) finds most of them for nothing; a $50-$150 shop smoke test finds the ones under the manifold that no ear will reach. The PCV clause: the PCV valve routes crankcase gases back into the intake, and when it sticks open it is a vacuum leak with a part number — a gradually developing rough idle, sometimes with oil consumption or a light oil mist around the engine. Its hose cracks on the same schedule and is the first thing to look at on a high-mileage four-cylinder with a lean code. Fix: replace the cracked or disconnected hose ($10-$40 DIY, 15 minutes), the PCV valve ($5-$20, a 10-minute DIY on most engines), a throttle-body gasket ($100-$200), or an intake gasket ($200-$400, a shop job).
The cheap cause first. A cracked PCV hose, a hose knocked off its nipple during somebody else's repair, or a torn intake boot is the cheap cause here, and all three are visible with a flashlight in five minutes. General Motors' own communication for a rough-idling truck with P0171 or P0174 stored sends the technician to exactly that part — the crankcase ventilation tube — before anything is bought.
Fix it yourself — the five-minute hose and PCV check. Engine off and cool. Follow every rubber hose leaving the intake manifold or throttle body and squeeze it along its length: one gone hard, glossy or cracked at the fitting is the part. Pull each hose off its nipple and look inside the end, because a split that only opens under vacuum hides while the hose is relaxed. Then pull the PCV valve and shake it — a working valve rattles, a stuck one is silent. Parts run $10-$40, the tool is your hands, and the worst outcome is that you have eliminated the cheapest suspect for nothing.
At the shop: "Rough only at idle, smooth on the highway, P0171 stored, and the idle hunts between about 800 and 1,300. I have already checked the PCV valve and the reachable vacuum hoses. Please smoke-test the intake and show me where the vapor comes out before quoting a gasket."
2. Dirty throttle body and idle air control — $15 to $150
Carbon from the PCV system builds on the throttle plate over years, and at idle — when the plate is barely open — even a thin layer restricts airflow enough to make the engine stumble. The visual check: pull the intake hose off the throttle body (two clamps, 30 seconds) and shine a light inside — shiny metal is clean, black carbon on the bore and plate edges is your cause. The AC-load test: does the roughness spike the moment the compressor kicks on? It adds a few horsepower of load, and a dirty throttle body can't open slightly to compensate. The idle air control clause: on cars built before drive-by-wire, idle is held by a separate valve that bypasses the throttle plate and carbons up the same way — the tell is an idle that sags and stalls with the AC on, or surges after a cold start, with a P0505 stored; two screws, same can of cleaner. The relearn clause: on drive-by-wire cars, a high or rough idle that began right after a battery disconnect or a throttle-body cleaning is a lost idle relearn, not a fault — a few cold-start-to-warm cycles, or the key-on sequence from the manual, and it settles. Fix: DIY cleaning with $15 throttle-body cleaner (engine off, on a rag, never sprayed into a running drive-by-wire engine); the idle air control valve cleaned or replaced on older cars; and the relearn done afterward so the fix doesn't look like a new problem.
The cheap cause first. A $15 can of throttle-body cleaner and a rag, and it is what a manufacturer tells its own technicians to try: a General Motors communication whose entire instruction, in NHTSA's summary, is to clean the throttle body if the engine stalls at idle when the air conditioning comes on. If your roughness spikes the moment the compressor engages, that sentence describes your car.
At the shop: "Rough at warm idle and it stumbles the instant the AC compressor engages. Please inspect the throttle bore for carbon and show me, clean it, and perform the idle relearn — and quote a throttle body only if cleaning does not hold."
3. Misfire at idle: worn plugs, a failing coil, or an uneven injector — $80 to $450
Each cylinder should fire consistently, and a misfiring one — from a worn plug, a bad coil, or an injector delivering less than its neighbors — contributes less per cycle, which at idle (600-900 RPM) is slow enough to feel as a distinct beat. Idle is where a misfire is most visible, because there is no road noise and no load to hide it. The cold-start tell: plug misfires are worst in the first few minutes and smooth out as the engine warms, because a cold chamber needs a strong spark that marginal plugs can't reliably deliver. The injector clause: an injector clogged with carbon sprays less than commanded, and at idle — minimal fuel demand, critical timing — even small flow variations create a beat with no temperature pattern that developed over months. A bottle of fuel-system cleaner ($15-$25) and about 100 mostly highway miles clears moderate deposits; a shop injector balance test names a weak one before anyone replaces the set. Fix: plugs as a full set (never just the one), the coil on a stubborn cylinder, cleaner first for a gradual beat and ultrasonic cleaning ($100-$200) if it doesn't hold, and replacement only for an injector with an electrical fault or physical damage — fuel injector replacement cost runs from a seal to four figures on a direct-injection engine. The same misfire felt on the road is jerking when accelerating, and one that only appears under load belongs to the sputtering page.
The cheap cause first. A set of plugs at $80-$200, which on many engines is maintenance you owed the car anyway. The expensive version of the same complaint is an engine that misfires because a valve, a ring or a timing component has moved — which is why the coil swap and a compression test come before anyone quotes internal work. One more cheap cause people skip: a stored misfire code a manufacturer has already answered with a software update, which costs nothing but the appointment.
Fix it yourself — the coil swap, step by step. Get the code read free and write it down exactly. A cylinder-specific code (P0301 through P0312) names the cylinder; P0300 means random or multiple, which points at something shared — fuel, a vacuum leak, or all the plugs worn together. For a single-cylinder code, unplug and unbolt that coil, move it to a neighboring cylinder, put that cylinder's coil where the first one was, clear the code, and drive until the light returns. Code followed the coil: buy one coil. Code stayed on the original cylinder: buy plugs, and buy the full set, because plugs wear together. Twenty minutes and no parts, and it stops you paying for both.
At the shop: "P0303 stored, worst in the first three minutes after a cold start. I swapped the coil to cylinder 2 and the misfire stayed on 3. Please confirm with a compression test on that cylinder before quoting anything past plugs, and check for a bulletin on this engine."
4. Engine mounts — $200 to $600 per mount
Engine mounts don't make the engine fire roughly — they fail to isolate its normal combustion vibration from the chassis, so each stroke transmits to the body instead of being absorbed. The hood check: with the engine running, have a helper briefly rev to 2,000 RPM and release — a healthy engine barely rocks, while failed mounts let it lurch forward under throttle and snap back. The hydraulic clause: many modern mounts are fluid-filled, and a leaking one shows a wet stain on its rubber and loses its damping long before it tears, so a mount that looks intact can still be empty. Fix: replacement with OEM or OEM-quality mounts, since cheap aftermarket ones often fail within a year and force the job to be redone. It is the one rough-running cause that throws no codes, which is exactly why the free Neutral test matters.
The cheap cause first. Before a set of mounts is quoted, the cheap cause is a mount that is merely out of position rather than worn out. Two manufacturers put that in writing. A Land Rover service message describes "poor idle quality" with "vibration at idle noticeably worse than comparison vehicles" and gives the cause as incorrect positioning of engine mounts following repairs in service or production. General Motors' preliminary information for the same complaint on full-size trucks tells the technician to try centering the engine and transmission mount first, and only then to shim the right mount. Neither is a parts sale. Ask which of your mounts has measurable play before you buy three.
At the shop: "Vibration in Drive with my foot on the brake, gone in Neutral, no codes stored. Please check whether the mounts are centered before quoting replacements, put the car on the lift with a helper loading the drivetrain, and show me which mount actually moves."
5. EGR valve stuck open — cleaning to a valve, priced by model
The EGR valve feeds a measured amount of exhaust back into the intake at cruise to lower combustion temperatures, and it is supposed to be shut at idle — exhaust in the idle mixture is inert gas taking the place of air, and the engine stumbles or stalls on it. Carbon holds the valve partly open, and on older vacuum-operated valves a leaking diaphragm does the same. The result is a rough idle that is perfectly smooth at cruise, where the valve belongs open anyway, and worst in the seconds after you come off the highway. The vacuum-pod test (free, older engines): with the engine idling, find the EGR valve on the intake or exhaust manifold and press the diaphragm's stem up with a finger, or pull its vacuum hose — if the idle changes, the valve is passing exhaust at idle. The code clue: P0402 (excessive flow) is this cause by name; P0401 (insufficient flow) is the opposite fault. On newer engines a scan tool can command the valve closed and watch the idle. Fix: remove the valve (two bolts on most) and clean the carbon out of it and its passage with the same throttle-body cleaner; replace it when cleaning doesn't hold or the diaphragm leaks. It is the idle cause most often mistaken for a vacuum leak — the difference is that a vacuum leak hunts and an EGR stumble arrives as the RPM falls.
The cheap cause first. Cleaning — and on some cars it is not even your bill. Lexus issued a warranty policy bulletin, POL16-02, for a rough idle and a rattle from the transaxle at cold start "due to a sticking exhaust gas recirculation (EGR) valve," with P0401 stored: a coverage document, not a repair quote. Toyota announced a warranty enhancement program for certain 2012-2013 Scion iQ vehicles with a rough idle "caused by internal carbon deposit in the combustion chambers." Before you pay to clean or replace an EGR valve, ask a dealer about coverage on your exact VIN.
At the shop: "It stumbles and nearly stalls as it comes down to idle after a highway run, then idles fine. P0402 stored. Please clean the EGR valve and its passage and re-test before quoting a valve, and check whether there is a warranty enhancement or policy bulletin for this engine."
A rough idle is the one symptom a parked recording can actually hear. Film the engine running, parked, for 30 seconds — the hunting needle, the beat of a misfire, the stumble as it settles — and upload it with an email; about ten minutes later a free report gives an Engine Health Score, a status for engine core, bearings, intake, belts, and exhaust, an independent second AI opinion, a likely cause with severity, a six-month outlook, and a repair-cost estimate — or says "inconclusive" if the clip can't support a verdict. It can't feel a mount (that's the Neutral test) and it doesn't read codes (that's the free scan); it's a second opinion before a shop sells a $750 tune-up.
Run the free tests in the right order
The five free checks are worth more in sequence than in isolation. Neutral first, because nothing else decides which half of the guide you are in. Then the scan — an empty scan is information too: no codes with a rough idle points back at mounts or a carboned throttle body. Then whichever test the code implies: hose-listen and the PCV rattle for a lean code, the coil swap for a cylinder-specific misfire, the EGR pod test for a stumble on the way down. In that order the sequence usually names the system, and often the part, before you have spent $15 on a can of cleaner.
Can engine mounts cause a rough running car? Yes, but not the way people assume — mounts don't make the engine run rough mechanically. A failed mount fails to isolate the engine's normal combustion vibration from the chassis: the engine fires normally, but that vibration transmits to the wheel, floor, and seats instead of being absorbed, so it feels like rough running. This is why the Neutral test is so valuable — it separates a mount problem from an engine problem for free. Shift to Neutral at a stoplight: if the roughness reduces significantly versus Drive, mounts are the cause, because Drive loads the drivetrain and strains the mounts while Neutral removes that load. If the roughness is the same in all gears, the cause is engine-side — not mounts. A helper revving with the hood open confirms it: failed mounts let the engine visibly lurch. A mount is felt, not heard — a PulsCar recording can't tell you this; the Neutral test can.
What 5,064 NHTSA rough-idle complaints show
Everything above comes from mechanics and physics. This section comes from a file nobody writing about rough idle seems to open: the complaints American owners file with the federal government, narratives included.
We loaded all 777,040 vehicle complaints received between 1 January 2015 and 8 September 2026 and kept the ones where a roughness word — rough, shake, vibrate, surge, hunt, fluctuate, misfire, sputter, stumble, judder, shudder — sits within sixty characters of an idle or stationary word in the same sentence. We threw out denials, "rough road", sudden-acceleration reports, and launch shudders (the car that shakes pulling away from a stop is a different symptom). That leaves 5,064 complaints describing an engine that runs rough while the car is standing still. We read three rounds of thirty random matches while tightening the query; the final round found four misfits in thirty, so treat every count here as carrying roughly one mis-sort in eight. The method file lists every filter and exclusion.
A rough idle still ends with a tow truck less often than the average complaint. In the rough-idle file, 3.89% report the vehicle was towed, 3.24% report a crash, and 0.85% report a fire. Across all 777,040 complaints the same fields read 5.11%, 4.37% and 1.82%. A hundred and thirty-seven injuries appear, and no deaths. Every one of those three is below the file's own baseline, and fire is less than half of it — which is the most useful reassurance on this page: the symptom that feels most alarming from the driver's seat is, in the federal record, one of the least dangerous. It is a money problem before it is a safety problem — which is why the free tests matter here more than the urgency does.
But one in five says it turned into a stall. 1,080 of the 5,064 narratives (21.3%) use the word stall, against 7.63% across all complaints. That is the boundary this page is really about. A rough idle that has never stalled is a repair you can schedule; a rough idle that has already died at a light is a stalling problem with a rough idle as its warning shot, and it belongs at the front of the queue. Two in ten crossing that line is not "most cases" — it is a minority — but it is a large enough minority that the honest advice is to fix a rough idle on a calendar, not on a whim.
It is not a high-mileage symptom. Median mileage is 70,000 miles (3,017 complaints report a usable figure), quartiles 32,000 and 107,000. The all-complaint median is also 70,000, so a rough idle arrives at exactly the fleet's average odometer. 27.5% are at or under 36,000 miles — the typical bumper-to-bumper limit — and 42.0% at or under 60,000, where a lot of powertrain coverage still lives. Before you authorize a dollar, check your in-service date: more than a quarter of the people who file over this symptom could have asked somebody else to pay.
Four percent were sent home without a diagnosis. 201 complaints (4.0%) contain a no-fault-found phrase — could not duplicate, unable to reproduce, no problem found, operating as designed, within specification. Across all complaints that family appears in 1.86%, so a rough idle is about twice as likely as the average complaint to end in a shrug. The practical answer is to arrive with conditions written down: cold or warm, in gear or in Neutral, with the air conditioning on or off, how many minutes after start, and what the tachometer does. A service adviser can work with that. "It idles rough" is what gets written up as no fault found.
What owners point at is not what fixes it. In the file, 653 narratives mention a misfire (12.9%), 396 name plugs or coils (7.8%), 182 the throttle body or idle air control (3.6%), 275 mention Neutral (5.4%), 162 the air conditioning (3.2%), 97 a cold start (1.9%) — and only 54 an engine mount (1.1%), 51 a vacuum leak or PCV (1.0%), 19 an EGR valve (0.4%). Read that as a map of vocabulary, not of causes: owners write down the part a shop said out loud. The two cheapest causes in this guide are named by almost nobody, which is exactly what you would expect of failures that get fixed quietly and never become a federal complaint.
Which models and model years dominate the rough-idle file
Read this as a map of where the noise is, not a reliability ranking. These are raw counts of unverified owner reports with no denominator: NHTSA does not publish how many of each model are on the road, so a best-seller out-counts a rare car even when it fails less often per vehicle. Publicity compounds it. Nothing in this table says any of these vehicles is unreliable, and no honest reading of complaint counts can say that.
What it does say is where the shape of the problem differs, and that part is real. One line is unlike every other line on this page: the Honda CR-V's median mileage at the complaint is 800 miles. Not eighty thousand. Eight hundred. Nine in ten of those cars had not reached 36,000 miles, and 187 of the 272 are a single model year. That is not wear — that is a vehicle that shook at a stoplight on the drive home from the dealership, and Honda published a bulletin about it. Set the CR-V aside and the file behaves as mechanics expect: Ford Escape at 98,000 miles, Chevrolet Malibu at 100,000, Pontiac G6 at 135,000, all consistent with plugs, carbon and hoses that have aged out.
Model years cluster at 2015 (394 complaints), 2013 (246), 2016 (214), 2014 (212) and 2012 (192) — a distribution shaped as much by how many cars of that age are still on the road and still under warranty as by anything mechanical. Full tables as CSV, with median mileage, warranty-window share, towed percentage and stall share for every make and for the top sixty models.
What owners said the repair turned out to be
1,457 of the 5,064 narratives contain both a repair verb — diagnosed, determined, replaced, installed, needed a new — and a named part family. Categorizing that text:
This is not a cause ranking, and reading it as one would cost you money. It is a ranking of which rough idles end up in a federal safety complaint. Nobody writes to the government because a $15 can of throttle-body cleaner fixed their stumble; people write when a four-figure engine bill lands, when a dealer says "operating as designed," or when the car stalls in traffic with a child aboard. That is exactly why engine internals sit at the top at 39.5% and a vacuum leak sits near the bottom at 7.7% — not because torn hoses are rare, but because a hose is a Tuesday and a short block is a grievance. Categories overlap, so the shares do not sum to 100%.
Used correctly the table still earns its keep, in three ways. Ignition and air metering together (229 mentions) outweigh engine internals, which says the middle of the price ladder catches most of the reports that name anything at all. Software alone resolved 51 cases — more than mounts and EGR combined — which is an argument for asking about bulletins before parts. And the battery and charging line at 74 is the surprise: a failing battery or a corroded ground can make a modern engine idle badly, because the control module is running on dirty voltage, and that is a $200 problem masquerading as a $2,000 one. Download the categories.
What the manufacturers' bulletins say — including the Neutral test, in GM's own words
The single most important claim on this page is the Neutral test: roughness that eases in Neutral is the mounts, roughness identical in every gear is the engine. It is fair to ask why you should believe that. Here is a manufacturer writing it down.
General Motors' preliminary information communication PIP5211, first issued 16 July 2014 and revised through PIP5211E and PIP5211F, tells technicians how to "diagnose and repair vehicles with a rough idle and/or vibration at idle in Drive," then adds the sentence this whole page is built on: "This condition will not be present in Park, or Neutral." The cause GM names is a ground-out within one or both engine mounts, and the document sends the technician to an oscilloscope to measure the disturbance rather than to a parts counter. PIP5137 (29 July 2013) covers the same complaint on full-size trucks and says to try centering the engine and transmission mount first, and only then to shim the right mount by 3 mm. Land Rover's SSM 72102 (12 May 2014) describes "poor idle quality" where "vibration at idle is noticeably worse than comparison vehicles" and blames "incorrect positioning of engine mounts following repairs in service or production."
Three manufacturers, one lesson: when a car is rough in gear and smooth in Neutral, the factory looks at mount position and mount condition — not at plugs, coils or injectors.
We searched every manufacturer communication NHTSA has published since 2015 whose summary names a rough idle, an idle vibration, a surging or hunting idle, an unstable or erratic idle, poor idle quality, a misfire at idle, a stall at idle or an engine that runs rough. The flat file carries one row per make, model and year, so 34,184 rows collapse to 966 distinct bulletins. Of those, 327 — 33.9% — mention software, reprogramming, a reflash, a calibration or a module update, and 29 mention a warranty extension, warranty enhancement, customer satisfaction program, special coverage or emissions recall.
The software share is not evenly distributed, and the spread is the finding:
- Stellantis: 302 idle bulletins, 223 of them mentioning software — 73.8%. For Chrysler, Dodge, Jeep and Ram, the factory answer to a rough idle is a flash roughly three times in four.
- General Motors: 201 bulletins, 16.4% software. GM's idle documents are overwhelmingly mechanical — mounts, throttle bodies, crankcase ventilation tubes.
- Ford: 85 bulletins, 15.3%. Toyota: 83 bulletins, 3.6%. Volkswagen Group: 34, 2.9%. Jaguar Land Rover: 48, 4.2%.
- Mercedes-Benz 56.5%, Nissan 52.4%, Subaru 46.7% sit in between, on much smaller counts.
The practical use is one question at the counter, and it is not confrontational: "Is there a technical service bulletin, a software update or a warranty extension for a rough idle on my engine?" A service adviser can answer from the manufacturer's own system in under a minute. On a Stellantis product the odds of a yes are genuinely good; on a Toyota they are poor, and you should expect a mechanical diagnosis instead. Neither answer is a verdict on the car — it is a fact about how that manufacturer writes its documents. Per-manufacturer counts as CSV.
Two more documents show how wide the cause list is. GM's PIP5341 (26 October 2015) covers a truck whose "engine runs and idles rough" with P0171 or P0174 stored; the fix is to inspect the positive crankcase ventilation tube for a vacuum leak, replace it, and sleeve it in conduit so it cannot chafe again — Cause 1, written by the manufacturer. GM's PIT3116B (13 May 2020) is shorter still: clean the throttle body if the engine stalls at idle when the air conditioning comes on. That is Cause 2 and its air-conditioning tell, in one sentence.
If your rough idle came with a code, the codes have a paper trail we have already counted: our Check Engine Code Index indexes manufacturer bulletins by code across the same NHTSA file, and the codes this page sends you to look for are all in it — P0300 with 431 bulletins from 14 manufacturer groups, P0301 with 276, P0171 with 105, P0401 with 59. Look yours up there before you buy the part it seems to name.
The cold rough idle has its own paper trail
Cause 3 says a rough idle worst in the first few minutes and gone once warm points at spark plugs. That stays the common case. But the cold-start rough idle is also the variant with the best odds of a factory answer that costs nothing, and three documents show why.
Acura service bulletin 18-063, dated 22 January 2019, is titled "2019 RDX: Engine Idles Rough on Cold Starts." Under SYMPTOM it says, in full: "The engine idles rough on cold starts." Under POSSIBLE CAUSES: "The engine idle is set too low." Under CORRECTIVE ACTION: "Update the PGM-FI software." The flat rate time is 0.2 hours — twelve minutes — and the normal warranty applies. Nothing is replaced. An owner who had been quoted plugs and coils for that car would have bought parts to fix a number in a calibration table.
Nissan bulletin NTB18-018 (1 March 2018) takes the opposite approach and defines the symptom so precisely that you can rule yourself in or out from the driveway: 2013-2015 Altima, "when the engine is started it runs rough (rough idle) for a few seconds, and then will idle normally," with the conditions given as an ambient temperature between 25°F and 45°F and an engine that has been off about six hours. That is a bulletin telling you which cold rough idle is theirs and which is yours.
Honda service bulletin 23-048 (21 July 2023) is the expensive end of the same symptom, and it is an emissions recall rather than a repair quote. Certain 2023 Accord Hybrids "may be equipped with an incorrectly machined engine block at the crankshaft position sensor (CKP) mounting location," and the bulletin lists the results as "a no start condition, low speed rough idle, engine RPM hunting," with DTC P0017 or P0339 stored. The corrective action is to inspect the block markings and, if necessary, replace the short block and the catalytic converter. Honda notes the expected failure rate is very low — but an owner of that car with a rough idle and a P0017 should not be paying for the diagnosis, let alone the block.
The lesson generalizes. A rough idle that is strictly cold, brief and repeatable is the kind of thing manufacturers write bulletins about, because it is measurable and it turns up on a service drive in winter. Ask for a VIN-level bulletin search before you authorize anything, and use the words cold start — a search on "rough idle" alone can miss a document filed under cold-start driveability.
Why a rough idle almost never becomes a recall
We ran the same symptom vocabulary against NHTSA's recall file: every campaign whose defect or consequence text names a misfire, rough running, a rough idle or a stall at idle. In the entire 2015-2026 file there are seven such campaigns, and only two are passenger vehicles — 18V-636 (Chrysler/FCA, 34,357 Alfa Romeo Giulia and Stelvio vehicles whose engines "may misfire causing the catalytic converter to overheat," remedied by an engine control module software update, free of charge) and 26V-493 (Genesis, 42 model-year 2027 G80s with a loose ignition coil ground bolt that can cause "an engine misfire or stall while driving," remedied by tightening the bolt). The rest are motorcycles, a scooter, a transit bus and medium-duty trucks.
That is not a hole in the search; it is the structure of the system, and knowing it changes what you ask for. A recall requires an unreasonable risk to motor vehicle safety, and a rough idle — the complaint file agrees, at one tow in a hundred — is usually a driveability problem. Notice what the two passenger-car campaigns share: both became recalls only once the misfire threatened something else, a converter hot enough to damage wiring in one case, a stall in traffic in the other. So do not wait for a recall on a rough idle; there almost certainly will not be one. The paperwork that helps you is one rung down — the bulletin, the warranty extension, the customer satisfaction program, the emissions recall — and unlike a safety recall, nobody mails you a letter about those. You have to ask.
Read your car's own rough running record in ten minutes
Every document quoted on this page is public and free, and the whole search takes about ten minutes.
- Find your VIN — the driver's-side windshield base, the door jamb sticker, your insurance card or your registration.
- Run it through NHTSA's recall lookup. It reports open, unrepaired safety recalls for that individual vehicle. A manufacturer must remedy a recalled defect without charge under 49 U.S.C. 30120, subject to the statute's own limits.
- Read the complaints for your exact year and model on the same site, searching the page for idle, rough, vibration and stall. Owners describing your symptom at roughly your mileage means a pattern, not a one-off.
- List the manufacturer communications for your model year. The titles are public even when the full text is not, and a title naming rough idle, idle vibration, idle quality or cold start is worth quoting by number.
- Ask the dealer two questions by phone before booking: is there any open campaign, warranty extension or customer satisfaction program for my VIN, and is there a bulletin for a rough idle on my engine? Write down the answer and who gave it.
- Then walk in with conditions, not adjectives. Cold or warm, in gear or Neutral, AC on or off, minutes after start, what the tachometer does, and any stored code. That is the difference between a diagnosis and a no-fault-found line on an invoice.
Three rough running traps, dissected
Trap one: the tune-up sold without the Neutral test. The situation: a car runs rough, a shop quotes plugs, coils and injectors, and the roughness continues, because the cause was failed engine mounts. What's real: the free Neutral test would have shown the roughness easing in Neutral — and GM's own bulletin says a vibration present in Drive but not in Park or Neutral is a mount. The defense: do the Neutral test, and ask the shop whether they did, before approving ignition or fuel work.
Trap two: both plug and coil replaced for one misfire. The situation: a P0301 misfire gets a new plug and a new coil, doubling the cost. What's real: the coil-swap test would have named which one, for free. The price: a $150-$400 coil bought needlessly, or vice versa. The defense: swap the coil to an adjacent cylinder first and see whether the misfire follows it.
Trap three: ignoring the flashing light. The situation: a rough-running car with a flashing light gets driven for weeks until the plugs are convenient to do. What's real: the active misfire was cooking the catalytic converter the whole time. The price: an $80 plug job becomes a $900-plus converter. The defense: a flashing light means stop and fix the misfire now — the converter is the expensive victim.
Three rough running quotes, decoded
These are composite scenarios, not individual customers — the amounts and outcomes are typical of the patterns above, assembled so the arithmetic is visible. We do not publish invented testimonials.
Scenario 1: the tune-up that was $400 of mounts. A shop replaced plugs, coils and injectors for a rough idle; it continued. The Neutral test, run afterward, showed it smoothing in Neutral, and $400 of mounts fixed it. Lesson: the free Neutral test comes before any tune-up.
Scenario 2: the $45 hose behind a "sick engine." A hunting idle with a P0171 was quoted a $750 tune-up. A listen along the intake found a torn intake boot; a $45 hose ended it. Lesson: idle-only roughness with a lean code is a vacuum leak — test before you buy parts.
Scenario 3: the $900 converter from a flashing light. Weeks of driving with a flashing light and a rough engine; the active misfire destroyed the converter. A $150 coil, caught early, would have prevented it. Lesson: a flashing light means stop now.
After the fix: keep the engine smooth
Replace spark plugs on schedule. Iridium plugs last 60,000-100,000 miles; overdue plugs are a leading cause of rough running and converter damage. Clean the throttle body every 30,000-50,000 miles — short-trip city driving carbons it up faster. Change the air filter every 15,000-20,000 miles. On a fuel-injected engine a clogged filter costs you power rather than mixture: the EPA and DOE's fueleconomy.gov states that replacing a clogged filter on a fuel-injected engine does not improve fuel economy but can improve acceleration, and the rich-running, plug-fouling story belongs to carburetors. Check the PCV valve at each service — the 30-second rattle test catches a stuck valve before it roughens the idle. And run Top Tier fuel, whose detergents cut the carbon that causes rough running over time. When a big repair is quoted, an independent shop and a second opinion beat a tune-up nobody diagnosed.
Should you buy a used car with a rough idle?
Sometimes, and at a discount you can calculate — which is more than you can say for most test-drive symptoms. The federal complaint file is unusually encouraging here: a rough idle tows fewer owners than the average complaint (3.89% against 5.11%) and crashes fewer (3.24% against 4.37%), so unlike a shudder at highway speed or a soft brake pedal, this is a negotiation rather than a walk-away. The work is in telling the cheap version from the expensive one before you sign.
Do the Neutral test on the test drive. Ten seconds at a light, and the seller cannot spin it. Roughness that halves in Neutral is mounts — a bounded job at $200-$600 per mount, priced from the model and deducted from the asking price. Roughness identical in every gear is engine-side, and that is where the range opens up.
Then scan it yourself. A $30 reader is the best money in used-car shopping. A stored P030X names a cylinder and points at $80-$400 of ignition; P0171 or P0174 with a hunting idle points at a hose. No codes at all is either a mount, a carboned throttle body, or a car whose codes were cleared in the parking lot an hour ago — which is why you check readiness monitors too: a full set of "not ready" monitors on a car with 60,000 miles means somebody cleared the memory recently.
Insist on a cold start. Ask the seller not to warm the car, and turn up early enough to touch the hood. The cold-only rough idle has the best odds of a factory answer and is also the variant a warmed-up car hides completely. A seller who already ran the engine for twenty minutes may be being kind; assume otherwise and come back.
Run the VIN before you negotiate, not after. Ten free minutes on NHTSA's site tells you whether this model and year has a documented idle problem, whether a bulletin or warranty extension exists, and whether a recall is still open. A documented, bulletined idle on a model with published fixes can be a genuine bargain: the fear is priced into the sticker and the repair is a known quantity.
The two cars to walk away from. One: a rough idle with a flashing check engine light, because the converter is being destroyed while you drive it and you cannot see how much has already gone. Two: a rough idle the seller describes as needing a tune-up, with no codes, no service history and no willingness to let an independent shop look at it. Everything else is arithmetic.
Your rough running plan: right now, today, this week
Right now (free):
- Do the Neutral test at a stoplight — smoother in Neutral means mounts, the same in all gears means engine-side.
- Get a free OBD scan at a parts store — P030X is a misfire, P0171 is lean, no code points at mounts or throttle body.
- If the light is flashing, stop driving — it's an active misfire cooking the converter.
Today (under $25): 4. For a lean code or hunting idle, do the hose-listen vacuum test and check the PCV valve. 5. For a stumble coming down to idle on an older engine, do the EGR vacuum-pod test; for a misfire code, do the coil-swap test.
This week: 6. Match the fix to the test — a hose or throttle-body clean for the hunting idle, plugs for cold-only, an EGR clean for the stumble on the way down, mounts if Neutral smooths it — and the sputtering page if it's rough at speed too. 7. Before any "tune-up" over $200 → confirm the shop did the Neutral test and has a specific code, not a guess. 8. Before buying both a plug and a coil → do the free coil-swap test first.
For the related guides: car sputtering under load (rough at speed — the other half of this symptom), stalling while driving, hesitation when accelerating, jerking when accelerating, shaking at idle, flashing light with shaking, and check engine light on. For parts and money: spark plug pricing, what a diagnostic should cost, and signs you're being overcharged. And our story explains why PulsCar exists.
Sources and documents
Primary documents and data, opened and read this month. Our own analysis is published as CSV so you can check it.
- General Motors PIP5211 / PIP5211E / PIP5211F, 16 July 2014 with revisions through 2017 — "rough idle and/or vibration at idle in Drive. This condition will not be present in Park, or Neutral," attributed to a ground-out within one or both engine mounts. Text as published in NHTSA's manufacturer communications file, records 10112114, 10080684 and 10099468; no scanned PDF exists for these records. File layout
- General Motors PIP5137, 29 July 2013 — rough idle or vibration at idle in gear on full-size trucks; center the engine and transmission mount, then shim the right mount 3 mm. NHTSA record 10113159.
- General Motors PIP5341, 26 October 2015 — engine runs and idles rough with P0171 and/or P0174; inspect and replace the positive crankcase ventilation tube, protected with conduit. NHTSA record 10114925.
- General Motors PIT3116B, 13 May 2020 — clean the throttle body if the engine stalls at idle when the air conditioning comes on. NHTSA record 10176314.
- Land Rover SSM 72102, 12 May 2014 — "poor idle quality... vibration at idle is noticeably worse than comparison vehicles," caused by incorrect positioning of engine mounts. NHTSA record 10103281.
- Honda service bulletin 15-046, "Vibration While Driving and/or Stopped in Gear," 2015 CR-V — vibrations felt through the driver's seat, the first listed condition being "while stopped with vehicle in gear"; the repair may include a software update that may affect fuel economy. NHTSA records 10100215 and 10108669.
- Acura service bulletin 18-063, 22 January 2019, 2019 RDX AWD — "Engine Idles Rough on Cold Starts"; possible cause, the engine idle is set too low; corrective action, update the PGM-FI software; 0.2 hours. PDF
- Honda service bulletin 23-048, 21 July 2023, 2023 Accord Hybrid emissions recall — an incorrectly machined block at the crankshaft position sensor mounting location causing "a no start condition, low speed rough idle, engine RPM hunting," with P0017 or P0339. PDF
- Nissan bulletin NTB18-018, 1 March 2018 — 2013-2015 Altima rough idle on start-up, defined between 25°F and 45°F after roughly six hours off. NHTSA record 10134857.
- Lexus warranty policy bulletin POL16-02, 16 June 2016 — rough idle and rattle at cold start from a sticking EGR valve with DTC P0401. NHTSA record 10091621. Toyota's warranty enhancement program for 2012-2013 Scion iQ rough idle from combustion-chamber carbon: NHTSA record 10131847.
- Recall 18V-636, Chrysler (FCA US LLC), 14 September 2018 — 34,357 Alfa Romeo Giulia and Stelvio vehicles that "may misfire causing the catalytic converter to overheat"; remedy, an engine control module software update, free of charge. NHTSA campaign record
- 40 CFR 86.1806-17 — the federal onboard diagnostics rule for model year 2017 and later vehicles: OBD systems must "detect malfunctions in the emission control system, store trouble codes corresponding to detected malfunctions, and alert operators appropriately," and engine misfire is listed among the major diagnostic monitors that may not be omitted. Section text
- 49 U.S.C. 30120 — a manufacturer must remedy a recalled defect without charge, subject to the statute's limits. Statute
- NHTSA complaint flat file — 777,040 vehicle complaints, 1 January 2015 to 8 September 2026. Field dictionary
- Our analysis, published — by make, top models, what owners said was repaired, idle bulletins by manufacturer and the method file. Licensed CC BY 4.0.
What we could not verify. There is no published, independent figure for how often a rough idle turns out to be each of the five causes; no such dataset exists, which is why this page ranks causes by pattern, mechanism and cost rather than by a percentage. We also could not re-read the repair estimators: RepairPal and the parts-store service pages refused our requests with an HTTP 403 on 18 September 2026, so every price range here carries its earlier verification date.
How these numbers were built: cross-checked against published 2026 repair-estimator ranges and typical independent-shop quotes (OBD scan free; MAF cleaner $8; throttle-body clean $15 DIY or $80-$150 shop; vacuum hose $10-$40 part or $40-$100 shop-fitted; PCV valve $5-$20 part plus $50-$100 labor if inaccessible; vacuum-leak smoke test $50-$150 (CarParts.com); spark plugs $80-$200; injector cleaning $100-$200, replacement $150-$450 each; ignition coil $150-$400; O2 sensor $150-$400; MAF replacement $200-$400; engine mount $200-$600 each; catalytic converter $900-$2,500; EGR valves priced by model and left unpriced here), at 2026 independent-shop labor rates. Prices reviewed quarterly — last verified July 2026.
Prices verified July 2026 and reviewed quarterly — no estimator would serve our checks this pass (RepairPal and the parts-store service pages returned HTTP 403 on 18 September 2026), so the ranges above are unchanged and should be treated as estimates you confirm at the counter. Technical documents verified September 2026: the manufacturer bulletins, the recall record and the regulation in the sources list were read on 18 September 2026, two of them as the original PDFs on static.nhtsa.gov and the rest as NHTSA's published record text, and the complaint and communication files behind our own analysis were queried the same day.
Have a rough-running pattern we didn't cover? Email [email protected] with whether it's smoother in Neutral, when it's roughest (cold, idle, or all speeds), and any OBD codes, and we'll add it to the next version of this guide.

