21 CFR Part 211 and its 5 linked instruments
How to read this page
- First column: 21 CFR Part 211. The others are every document linked to it.
- Scroll sideways if there are more columns than fit.
- It does not compare content. It does not say which is stricter. Read the official texts for that.
| 21 CFR Part 211 | FDA CPG 425 series | FDA quality systems guide | FDA data integrity Q&A | FDA AI guidance (drugs) | FDA AI warning letter | |
|---|---|---|---|---|---|---|
| Authority | FDA | FDA | FDA | FDA | FDA | FDA |
| Legal weight | Law & regulation | Guidance & Q&A | Guidance & Q&A | Guidance & Q&A | Guidance & Q&A | Warning letter: a flag, not a rule |
| Status | In force | Final | Final | Final | Draft | Warning letter |
| Last verified | Not yet | Not yet | Not yet | Not yet | Not yet | Not yet |
| Summary | US CGMP rules for finished pharmaceuticals. Section 211.68 allows computers and automated equipment in manufacture if they are routinely checked under a written program. It requires controls so only authorised people change master records, checks on input and output, and a backup of data. | FDA compliance policy on computers in drug manufacturing. Under CGMP, hardware is generally treated as equipment and applications software as records, and the system's use decides which sections apply. The series also covers vendor responsibility, source code for process control, input and output checking, and identifying persons on batch records. | FDA guidance to help manufacturers use modern quality systems and risk management to meet the CGMP regulations in 21 CFR parts 210 and 211. | FDA's question-and-answer guidance on data integrity in drug manufacturing: ALCOA, audit trail review, access controls, shared logins and when electronic data are CGMP records. | FDA draft guidance proposing a risk-based credibility framework for AI models whose output supports regulatory decisions about drugs and biologics. Still a draft (docket FDA-2024-D-4689). | Described as FDA's first CGMP enforcement on AI misuse: the firm used AI to generate specifications and procedures without adequate review. FDA said AI-generated documents must be reviewed for accuracy. |
| Lifecycle |
|
|
|
|
|
|
| Source | Official text Source: FDA. Public domain. | Official text Source: FDA. Public domain. | Official text Source: FDA. Public domain. | Official text Source: FDA. Public domain. | Official text Source: FDA. Public domain. | Official text Source: FDA. Public domain. |
These facts are placed side by side from each record. This page does not say which is stricter or whether requirements differ. Read the official texts.