Current download snapshot: Sep 28, 2026. The library contains 3,677 model-year dossiers for selected US-market makes, covering model years 2000–2025. The snapshot date is when files were retrieved; it is not a guarantee that every underlying record was updated on that date.
The original sources.
| Dataset | How it is used |
|---|---|
| Owner complaints | Distinct ODI numbers per model-year. |
| Safety recalls, before 2010 | Distinct campaign numbers per model-year. |
| Safety recalls, 2010 onward | Campaign concern, remedy and applicability fields. |
| Investigations | Distinct investigation IDs per model-year. |
| Manufacturer communications | All seven received-year archives from 1995 through 2026; distinct NHTSA IDs. |
Field dictionaries accompany each download. File size, download timestamp, source URL, server last-modified value (when present) and a SHA-256 checksum are retained in the local source manifest. No live API is needed when someone opens a dossier.
Bulk snapshots and live lookups can differ.
The displayed totals come from the downloaded bulk archives, not from a live API response. During a September 28, 2026 spot-check, the Toyota RAV4 and Honda CR-V samples matched the live complaint API. For the 2018 Ford F-150, the bulk archive contained 2,308 distinct ODI IDs while the live endpoint returned 75. An independent recount of the downloaded file confirmed the larger bulk total; the reason for the endpoint difference was not established. Follow the bulk source and record identifiers when reconciling counts.
What qualifies for a dossier.
A model-year must have at least 50 distinct vehicle complaint IDs, plus either five manufacturer communications, one recall campaign or one investigation. A model family must have at least two qualifying years. This is a publication-depth rule, not a measure of safety or a definition of a “problem car.”
Coverage is deliberately selective. Missing years may have too few records, inconsistent source naming or fall outside the selected makes and year range. Missing from this site does not mean free of problems.
How records are joined.
We join the sources by make, model and model year. Whitespace and capitalization are normalized, with explicit aliases for F150/F-150, F250/F-250, F350/F-350, CRV/CR-V, HRV/HR-V and RAV-4/RAV4. Other source model names remain separate; we do not silently merge trims or assume shared engines.
Complaint rows are grouped by ODI number within each vehicle. Communications are grouped by NHTSA communication ID, recalls by campaign number, and investigations by investigation ID. Repeated component rows do not inflate each vehicle’s headline total. The same official document may legitimately appear on several vehicle dossiers.
Component groups and charts.
Original NHTSA component labels are mapped into broad groups such as engine, electrical system and transmission/drivetrain. Power-train labels belong to the broad drivetrain group; they do not establish a transmission-specific diagnosis. Labels without a supported match appear under other components. Original labels remain available in each dossier.
A complaint can belong to several groups. Each category’s share uses distinct complaints for that model-year as its denominator. The timeline uses the date the report was received and shows the latest 15 years containing reports. Missing dates are not imputed. Mileage summaries use supplied positive values up to one million miles and identify the number of usable reports.
Original-document explainers.
Selected warranty and repair programs have a separate document-reviewed explainer. These pages identify the original PDF, its download checksum, the pages supporting the facts and the review date. We summarize the manufacturer’s stated concern, response, coverage and exclusions in our own words. We do not republish the full PDF or its illustrations.
A review is tied to a specific document version; it is not a live VIN check or assurance that no later revision exists. Expired deadlines and customer cost-sharing remain visible. If the PDF disagrees with its bulk-file summary, we identify that discrepancy. Records without a reviewed explainer remain source summaries, not verified coverage decisions.
What we do not infer.
- No failure probability, safety score or “best year” ranking: exposure and sales denominators are not supplied consistently.
- No causal match between a complaint and a bulletin solely because they share a component.
- No assurance that a bulletin covers every trim, engine or VIN in the model-year.
- No assurance that a recall remains open for an individual car. Use the current VIN lookup and manufacturer records.
- No automatic claim that one bulletin supersedes another. The source replacement field is deprecated for newer records.
Rights and personal information.
NHTSA’s terms of use describe the information on its site as public information that may be copied or distributed. We attribute the agency and retain identifiers so records can be checked. That is not a blanket license claim for every manufacturer PDF, logo, photograph or third-party work.
Recallfolio republishes structured record summaries and original aggregate calculations. It does not mirror manufacturer PDFs or use manufacturer logos. Complaint narratives, VINs, owner names, addresses, telephone numbers and city-level complaint details are excluded from the public output. The homepage preview uses actual vehicle records. Social cards use the site’s typography, vector graphics and record counts.
Updates and corrections.
This is a fixed, dated edition. Downloads and rebuilds are repeatable but no automatic refresh schedule is active. Source snapshots are retained; unchanged transformed records keep their prior content-update timestamps. A refresh must pass validation and a successful build before replacing a published edition.
Before parsing, the importer verifies source byte counts and SHA-256 checksums against the manifest. It stops when a source dictionary differs from the reviewed version and rejects rows shorter than the expected format. These guards do not guarantee detection of undocumented upstream field changes. Refresh reviews identify added or removed vehicles and substantial record-count decreases before publication. Report a correction with a page URL and source identifier. Nothing here replaces the agency or manufacturer’s current record.