Written and maintained by CASRAI Editorial Board
Published · Last updated
On this page
The FDA complete response letter database (often searched as the “FDA CRL database”) is a public collection of complete response letters that the U.S. Food and Drug Administration has sent to drug and biologic sponsors, published through the openFDA project. You can search it by company, approval status and year on a web page, query it as JSON through the openFDA API, or download the letters as PDF archives. It is free to use and needs no registration or eligibility check. As described on the official openFDA pages on 2026-11-12, it holds letters for both approved and unapproved New Drug Applications (NDAs) and Biologics License Applications (BLAs).
What it is and who runs it
The database is part of the openFDA transparency tools, run by the FDA. The openFDA “Complete Response Letters” page states that the collection contains complete response letters (CRLs) for approved and unapproved NDAs and BLAs. A CRL is the letter the FDA sends when it will not approve an application in its present form. Under 21 CFR 314.110, the letter describes the specific deficiencies the agency found, and the regulation says it reflects a complete review of the data submitted. For the concept itself, including what a sponsor can do next, see our guide to the FDA complete response letter (CRL) and the dictionary entry for complete response letter. This page is about the database, not the regulatory pathway.
The CRL database is a separate tool from the other places FDA publishes regulatory records. It is not the same as Drugs@FDA, which publishes approval packages, and it is not the inspection data covered in our guide to the FDA inspection database. It also sits inside the wider openFDA service described in our openFDA guide.
The “FDA CRL database” search variant
People searching for “FDA CRL database” are looking for the same resource. The official entry points are the search page at open.fda.gov/crltable/ and the API documentation page at open.fda.gov/apis/transparency/completeresponseletters/. The FDA-hosted pages are the ones that state the scope, redaction rules and legal basis, so they are the ones to cite.
Who can use it (eligibility)
There is no eligibility requirement. The collection is public, and the openFDA API serves it without a sponsor account. The openFDA response carries its own terms link and a data-use disclaimer: users are told not to rely on openFDA to make decisions about medical care and to assume that all results are unvalidated. The API provider may also restrict access in line with its Terms of Service. The licence and terms are linked from every API response (open.fda.gov/terms/ and open.fda.gov/license/).
In practice the audience is regulatory affairs staff, clinical research professionals, analysts and academics who want primary-source examples of how the FDA states its deficiencies. The data supports reading and text analysis of past letters. It does not tell you whether a given application will receive a CRL, and it is not a substitute for advice from regulatory counsel.
What it contains
The API page describes each record as a CRL that openFDA converted to JSON from the original PDF. Based on a live query of the endpoint (api.fda.gov/transparency/crl.json) read on 2026-11-12, a record carries these fields: letter_date, letter_year, approval_status, approver_name, approver_title, approver_center (a list of FDA offices and divisions), company_name, company_rep, company_address, application_number (a list, for example in the form “NDA 210649”), file_name, letter_type and text (the full text of the letter). The official page notes that openFDA may change some field names, so check a live response before building anything that depends on a field.
Approved and unapproved applications
The approval_status field separates letters for applications that were later approved from letters for applications that were not. The two download archives follow the same split: approved_CRLs.zip and unapproved_CRLs.zip, both hosted at download.open.fda.gov and containing the letters as PDFs. At the API’s last update (meta.last_updated 2026-08-13, read on 2026-11-12), the endpoint reported 458 records, of which 309 were marked Approved and 149 Unapproved.
Not every record is a complete response letter
This is the point most summaries miss. The letter_type field shows that the collection includes other FDA letter categories. In the same reading, 445 records were typed COMPLETE RESPONSE and 8 were typed as other letters: TENTATIVE APPROVAL (4), PROVISIONAL DETERMINATION (1), CORRECTED PROVISIONAL DETERMINATION (1), REFUSAL TO FILE (1) and RESCIND COMPLETE RESPONSE (1). Five of the 458 records carry no letter_type value, so the letter_type counts sum to 453. Filter on letter_type before you count “CRLs”, otherwise your total will include letters that are not CRLs.
Years covered
The “time period covered” line on the API page reads 2020 to 2024, but the live data is broader: letter years in the endpoint range from 2002 to 2026, and the largest yearly counts at the reading date were 2024 (69 records) and 2025 (59 records). Counts for early years are very small (one or two records a year for 2002 to 2011), so the database should not be treated as a complete historical record of everything the FDA has ever sent. The data table below gives the full breakdown by approval status, letter type and year.
Redactions
The FDA redacts confidential commercial information and trade secret information before release and assesses each CRL independently. The official page cites the Trade Secrets Act (18 U.S.C. 1905) and section 301(j) of the Federal Food, Drug, and Cosmetic Act (21 U.S.C. 331(j)). It also lists these legal authorities for the disclosure program: the Freedom of Information Act (5 U.S.C. 552(a)), FDCA section 505(l) (21 U.S.C. 355(l)), 21 CFR Part 20, and 21 CFR 312.130, 314.430 and 601.51. Expect blacked-out passages in individual letters; a redaction does not mean the text was lost in conversion.
How to search and download
There are three routes, and they suit different tasks.
- Web search page (open.fda.gov/crltable/). Enter a company name (the page uses *** to mean all companies), choose a status of All, Approved or Unapproved, and choose a year. This is the quickest way to look up one sponsor or one year.
- openFDA API (api.fda.gov/transparency/crl.json). Returns JSON with the fields listed above and supports the standard openFDA
search,countandlimitparameters. Acount=approval_statusorcount=letter_typequery returns the facet totals shown in this page’s data table. Counting on the field name without the.exactsuffix was what worked in our reading; the.exactform returned an error on this endpoint. - Bulk PDF archives. approved_CRLs.zip and unapproved_CRLs.zip contain the original letters. Use these when you need the document as issued rather than the OCR text field.
The text field makes full-text searching possible, for example to find letters that mention a given application number or a particular category of deficiency. Because that text comes from converting PDFs, spot-check anything you quote against the PDF in the archive.
Update cadence and timeline phases
The API page lists the update frequency as “Infrequently”. The FDA states that it will keep adding CRLs as they are issued, and that it may also release batches drawn from archives for withdrawn and approved applications. For you, that means the database lags real-world events. A sponsor announcing a CRL does not mean the letter is in the database yet, and a letter in the database may be years old. The official pages describe the steps only in outline: the FDA redacts each letter after an individual assessment, and openFDA converts the released PDFs to JSON. They state no fixed time between issuing a letter and publishing it, so do not assume one.
A sample record read on 2026-11-12 stated that the sponsor should resubmit or take other action within one year, in line with 21 CFR 314.110, and that otherwise the FDA may consider the application withdrawn under 21 CFR 314.65. Treat that as the wording of one letter, and read 21 CFR 314.110 for the rule itself.
Common mistakes
- Treating the total as the number of CRLs ever issued. The database is a published collection with an “infrequent” update schedule, not a complete register. Always quote the
last_updateddate with any count. - Counting every record as a CRL. Filter
letter_typeto COMPLETE RESPONSE first. - Trusting the “2020 to 2024” label. The data includes older and newer letters. Use
letter_yearinstead. - Using it for medical decisions. The openFDA disclaimer says to assume all results are unvalidated.
- Ignoring redactions and OCR error. Compare quotations with the PDF before citing a letter in a regulatory briefing.
- Using it to predict outcomes. A published letter shows what the FDA said about one application. It does not give the base rate of CRLs among all submissions, because the denominator is not in the data.
Related databases
| Resource | What it holds | Where to read more |
|---|---|---|
| openFDA CRL database | Published complete response and related action letters for NDAs and BLAs, with full text and PDF archives | open.fda.gov/crltable/ |
| openFDA more broadly | The API service that hosts the CRL endpoint | openFDA guide |
| FDA inspection data | A different FDA transparency dataset, covering inspections rather than application decisions | FDA inspection database |
| CRL concept and rules | What a CRL is, what it means for a sponsor | FDA complete response letter guide |
| Regulatory intelligence | Using public regulator sources as a function inside an organisation | Regulatory intelligence function |
FAQ
What is the FDA complete response letter database?
It is the FDA’s public collection of complete response letters for approved and unapproved NDAs and BLAs, published through openFDA and available as a search page, a JSON API and PDF archives.
Is the FDA CRL database free and public?
Yes. The search page, the API and the downloads are open, subject to the openFDA terms and licence linked from the API responses.
Does it include rejected (unapproved) applications?
Yes. The official page says it covers both approved and unapproved applications, and the approval_status field and the two download archives keep them apart.
How often is it updated?
The official page lists the update frequency as “Infrequently”, and says the FDA will keep adding CRLs as issued and may release archive batches.
Are the letters redacted?
Yes. Confidential commercial information and trade secret information are redacted before release.
Can I rely on it for a regulatory or medical decision?
No. openFDA states that results are unvalidated and should not be relied on for medical care decisions. Check any quoted passage against the PDF.
Sources
- FDA openFDA, Complete Response Letters API page, https://open.fda.gov/apis/transparency/completeresponseletters/ (read 2026-11-12).
- FDA openFDA, CRL search page, https://open.fda.gov/crltable/ (read 2026-11-12).
- openFDA transparency API endpoint, https://api.fda.gov/transparency/crl.json (queried 2026-11-12; meta.last_updated 2026-08-13).
- 21 CFR 314.110, Complete response letter to the applicant, https://www.law.cornell.edu/cfr/text/21/314.110 (read 2026-11-12).
Data: openFDA CRL database record composition by approval status, letter type and letter year
Record counts from the openFDA transparency CRL endpoint, compiled from count queries on the approval_status, letter_type and letter_year fields. The API reported 458 total records (one record has no letter year, so year counts sum to 457). Five records carry no letter_type value, so letter type counts sum to 453. Counts change as the FDA adds letters.
Period covered: 2002/2026. Coverage: United States.
| Facet | Value | Records (records) |
|---|---|---|
| Approval status | Approved | 309 |
| Approval status | Unapproved | 149 |
| Letter type | COMPLETE RESPONSE | 445 |
| Letter type | TENTATIVE APPROVAL | 4 |
| Letter type | PROVISIONAL DETERMINATION | 1 |
| Letter type | CORRECTED PROVISIONAL DETERMINATION | 1 |
| Letter type | REFUSAL TO FILE | 1 |
| Letter type | RESCIND COMPLETE RESPONSE | 1 |
| Letter year | 2026 | 32 |
| Letter year | 2025 | 59 |
| Letter year | 2024 | 69 |
| Letter year | 2023 | 30 |
| Letter year | 2022 | 25 |
| Letter year | 2021 | 48 |
| Letter year | 2020 | 35 |
| Letter year | 2019 | 38 |
| Letter year | 2018 | 33 |
| Letter year | 2017 | 25 |
| Letter year | 2016 | 20 |
| Letter year | 2015 | 9 |
| Letter year | 2014 | 8 |
| Letter year | 2013 | 12 |
| Letter year | 2012 | 4 |
| Letter year | 2011 | 2 |
| Letter year | 2010 | 2 |
| Letter year | 2009 | 2 |
| Letter year | 2008 | 1 |
| Letter year | 2007 | 1 |
| Letter year | 2004 | 1 |
| Letter year | 2002 | 1 |
Sources
- openFDA transparency API, crl.json (count queries on approval_status, letter_type, letter_year; meta.last_updated 2026-08-13) (accessed 2026-11-12)
- openFDA Complete Response Letters API page (accessed 2026-11-12)
Compiled by CASRAI from the sources above. Free to reuse under CC BY 4.0 with attribution.







