PULSCAR ANALYSIS — P0303 (CYLINDER 3 MISFIRE DETECTED) IN NHTSA'S PUBLIC FILES PulsCar analysis of NHTSA public data, September 2026. License: CC BY 4.0. Attribution: "PulsCar analysis of NHTSA public data, September 2026." Published alongside https://www.pulscar.io/blog/p0303-code ------------------------------------------------------------------------------ Prepared September 2026. Source data: NHTSA public flat files (complaints received 2015 to 21 September 2026, 780,071 records; manufacturer communications from January 2015) and documents downloaded from static.nhtsa.gov. 1. Complaints (1,301, all read in full, two readers) First search (1,029): every complaint whose narrative names P0303 in any spelling (P0303, PO303, P 0303, P-0303, P030300, P0303-00, "code/dtc/cel ... 303") — 391; a code range that contains P0303 written as a span ("P0301-P0304", "P0301 through P0304", "P0302 -P0306") — 45; or a misfire on cylinder 3 in words ("cylinder 3 misfire", "misfire on #3", "3rd/third cylinder misfire", "number 3 coil/injector/plug ... misfire") — 593. Completeness check (272 added): before writing, an independent searcher ran 26 patterns over every narrative not already in the pool — P303, p303, 0303 and "303 code" without the zero or the P; P030X wildcards; ranges starting at P0300 or written in words; cylinder lists containing 3 ("cylinders 2 and 3", "1-3", "#2, #3 and #4"); "a code for cylinder 3"; coolant, compression or oil findings on cylinder 3 with a misfire. 1,740 candidates matched; 365 were read in full; 272 were kept and 95 rejected with reasons (audit_report in codebook.txt, section 3). The found_by column says which search found each complaint, and audit_phrase gives the wording that matched. Coding: the codebook (codebook.txt) was written, dated and hashed before any complaint was read. Two independent classification passes coded every complaint separately, each from the codebook only, without seeing the other's codes: one category (C coolant; M mechanical; F fuel or air; I ignition; E electronic or software; U unresolved or not stated; N not relevant) and four flags (cyl3_confirmed, multi_cylinder, parts_before_diagnosis, stall_or_power_loss). Passes: two independent passes for the first 1,029 complaints and two further independent passes for the 272 audit additions. Agreement over all 1,301: category 1,185 (Cohen's kappa 0.87); cyl3_confirmed 1,283 (0.87); multi_cylinder 1,250 (0.91); parts_before_diagnosis 1,250 (0.91); stall_or_power_loss 1,266 (0.94). Every disagreement (244 complaints with at least one differing field) was read in full and resolved under the written adjudication rules R1 to R11 (codebook.txt); a rule sweep of the agreed codes then corrected 4 more. The complaints file carries both passes' original codes, the final code, the resolution and the rule applied. Final: 1,286 relevant, 15 not relevant. C 562; U 316; M 291; F 67; I 32; E 18. cyl3_confirmed yes 1,226; multi_cylinder yes 402; parts_before_diagnosis yes 472; stall_or_power_loss yes 468. Ford 553, Honda 209, Hyundai 100, Chrysler 87, Kia 83, Chevrolet 36, Jeep 36, Dodge 35, Lincoln 30 and others. Median model year 2017; median reported mileage 90,000 (415 complaints give a plausible one; 2 readings above 1,000,000 miles excluded). By year received: 2015 14, 2016 16, 2017 22, 2018 26, 2019 49, 2020 73, 2021 49, 2022 98, 2023 191, 2024 258, 2025 262, 2026 (to 21 September) 228. Figures count complaints, not vehicles. Nothing here is a failure rate. 2. Owner-reported amounts (303) Every dollar figure an owner reports for a diagnosis or test, an ignition, fuel or air part, a cooling system or head gasket repair, a valvetrain repair, or an engine repair, rebuild or replacement on the owner's car in the story of its cylinder 3 misfire, in p0303-owner-reported-amounts.csv, with what it was for, whether the narrative says it was paid, billed, quoted, offered, covered or not stated, a group and a make group. Excluded (each decision recorded with its reason): vehicle prices, loans and payments, trade-in and buy-back offers, rentals and towing, repairs of unrelated systems, figures attributed to other people or the owner's own research, totals that mix unrelated work, and an earlier estimate when the final bill for the same job is given. Rows for complaints coded N are dropped. Of 303 figures, 183 are quotes and 88 are described as paid. In the article, engine statistics leave out one $375 contribution a dealership offered (164 engine figures, median $8,396.50). 3. Manufacturer documents (745 distinct, read in full) First search: every manufacturer communication whose NHTSA index summary names P0303 in any spelling, a range containing it, a cylinder 3 misfire, or companion codes for cylinder 3 (P0203, P0353, P1064, P349A): 665 NHTSA records; 595 distinct documents read in full; 66 byte-identical copies marked as duplicates; 4 records with no PDF on NHTSA. Completeness check: an independent search of the index for P030X wildcards, "No. 3 cylinder", "#3 cylinder", "third cylinder", cylinder pairs containing 3, P0303:00, and the MultiAir cylinder 3 codes P106A and P104A found 176 more records; 150 new distinct documents were read in full (25 byte-identical copies, 1 without a PDF): 745 in all. Near-identical documents (revisions and model-by-model copies) were read by comparing each against a fully read representative, sentence by sentence, and opening the full file wherever they differed. Empty or scrambled text layers were rendered to images and read. One document (Honda 26-004, 11036403) had a truncated first download; it was downloaded again and read in full. Of the 745, 680 are relevant and 65 are not (among them Infiniti documents where "P0303" is an internal campaign ID for a speedometer recalibration, and Jeep documents about injector circuit code P02F0). 461 relevant documents print P0303 itself (P0303, P030300, P0303-00 or P0303:00) in their own text; the rest carry it inside a range or a P030X wildcard, or concern a companion code. The column cylinder_3_specificity records whether the document singles cylinder 3 out (SINGLED OUT: cylinder 3, or a pair or bank containing it, named apart from the others — this also includes MultiAir P106A/P104A circuit codes, which are not misfire codes), lists it in the family (FAMILY LIST), or does not name it (NOT NAMED). Nine document families single out a misfire on cylinder 3: Mazda 01-010/18, Land Rover SSM74161 and Jaguar SSM74160, Audi 2060491/1, Ford 22-2017 and 24-2257, BMW SI B12 10 16, MINI SI M12 11 14, Infiniti ITB21-009, Lexus L-SB-0008-21, and Honda's piston ring bulletins (13-078 to 13-082, 25-060, 25-061, 26-001, 26-002 and owner letter ONL6Z6072025), which repair cylinders 1 to 3 when P0301, P0302 or P0303 is stored without P0304. Chrysler X56 (09-002-14 REV. B, 10139897) is cited in the article from the companion P0302 study, where it was read in full; it does not name P0303. Every verbatim quote in the article was checked by script against the document or complaint text. Coverage status in September 2026 is our computation from the terms printed in each document, not a statement by the manufacturer. Limits: complaints over-represent engine failures and denied claims; owners report the diagnosis as they understood it; searches of NHTSA's index cannot see bulletins whose summary is blank or codes that appear only inside a PDF. Nothing here is a failure rate.