Methodology & editorial policy
How RxRecalls is built: where every record comes from, what we add to it, what we never change, and how to get an error fixed.
Sources
We publish only official FDA data. Nothing on this site comes from press coverage, social media or reader reports.
- Recall records come from the FDA drug enforcement reports, published through openFDA. Each recall on this site is one record in that dataset, with the FDA's recall number, classification, status, recalling firm, product description and reason.
- Recall announcements. The serious recalls are usually announced by the recalling firm, and FDA posts the announcement on its Recalls, Market Withdrawals & Safety Alerts page the same or the next day — weeks before the recall is classified and reaches the enforcement reports. We list those announcements (drugs only) with their date, firm, product, reason and NDC codes, marked “not yet classified”, and link to FDA’s notice for the lot numbers. We do not copy the firm’s press release. Once the same recall appears in the enforcement data, the announcement points to it.
- Product details (brand and generic name, strength, dosage form, labeler) come from the FDA National Drug Code (NDC) directory, also via openFDA, matched by the NDC codes that appear in the recall.
openFDA data is released into the public domain (CC0). This site is independent: it is not affiliated with or endorsed by the FDA.
How often it is updated
We check FDA’s recall announcements every hour, and openFDA for new and changed recalls every day, and rebuild the complete dataset, together with the NDC directory, once a month so that corrections FDA makes to older records are picked up too.
FDA itself publishes enforcement reports weekly, and a recall usually appears there a few weeks after the firm starts it, once FDA has reviewed and classified it. So the newest recall in the register is normally older than the date we last synchronised. Right now the register holds 18,003 recalls; the most recent was initiated Sep 15, 2026, and the last sync with openFDA was Oct 10, 2026. Treat this site as a reference, not a real-time alert service.
What we add — and what we never change
The FDA's own text is shown unchanged on every recall page: the product description, the reason for the recall, the classification and the status. We do not rewrite, summarise or soften it. On top of it we add a few things to make the records easier to search:
- Cause type. Each recall is sorted into one of 16 cause types (for example sterility, labeling mix-ups, impurities) by fixed text rules applied to the FDA's reason. Recalls the rules cannot place go under “Other” (currently about 2% of records). The original reason is always printed next to the cause type, so you can judge the match yourself. See recall causes.
- Drug names. A recall is listed on a drug's page when that drug's name appears as a whole word in the FDA product description. Brand and generic names are not merged automatically, so it is worth searching both.
- NDC codes are extracted from the product description and linked to the NDC directory, which is the most precise way to check a specific package.
- Firm names are normalised (for example “Inc.” vs “Incorporated”) only to group a firm's recalls together; the firm name shown is the one in the FDA record.
- Recalls not yet classified are shown as “pending classification” and identified by FDA's event ID until FDA assigns a recall number. When it does, the pending entry is replaced by the numbered one.
- Plain-language guidance on what each risk class means for you is general and identical for every recall of that class. It is not tailored medical advice.
Accuracy checks
- Imports only add or update records. If a download from openFDA fails, or the NDC directory arrives incomplete, the run stops and the data already online stays as it was.
- Counts and statistics on the site (number of recalls per drug, cause, year or class) are computed from the stored records when the page is built, not typed by hand, so they cannot drift from the data.
- Every recall page shows the FDA recall number and links to FDA’s enforcement report database, so any record can be checked against the original.
Use of AI
Recall records, classifications and numbers on this site are never written or altered by AI. Some explanatory texts — for example the descriptions of recall causes and the guide on how to check a medication — were drafted with the help of an AI writing tool and then checked against the FDA data and FDA's published definitions before publication.
Editorial independence
- The site carries no advertising and no sponsored content, and no firm can pay to have a recall added, removed or reworded.
- We do not decide whether a product is recalled or how serious it is — FDA and the recalling firm do. We report their decisions.
- This is not medical advice. Never stop a prescribed medication based on this site alone; talk to your pharmacist or doctor first.
Corrections
If something on this site does not match the FDA record, or a recall is listed under the wrong drug or cause, please tell us via the contact page, with the link to the page. Errors in our own processing (matching, cause type, display) are fixed at the source so the fix applies to every affected page. Errors in the FDA record itself are FDA's to correct; once FDA updates the record, the change reaches this site with the next sync.
What the risk classes mean
- Class I Reasonable probability that use will cause serious harm or death.
- Class II May cause temporary or medically reversible harm; serious harm unlikely.
- Class III Unlikely to cause harm, but the product violates FDA regulations.
Limitations
- A recall usually affects specific lot numbers, not every package of a drug. Always compare the lot number on your package with the recall.
- The register covers drug recalls in FDA's enforcement reports. Safety alerts, shortage notices and other FDA actions that never became a recall are not included.
- Older records can be incomplete where the original FDA record is incomplete.