What must a GMP laboratory record contain for each test?

Why

21 CFR 211.194 lists what a laboratory record must hold so that someone else could reconstruct the test. It starts with the sample: where it came from, quantity, lot, date sampled and date received. Then the method, with a reference to the data showing it is accurate and reliable, and for compendial methods a verification that it works under actual conditions of use.

Then the amount of sample, the complete record of all data including graphs, charts and spectra, all calculations with units, conversion and equivalency factors, and the result compared with the standard. The analyst initials or signs and dates it, and a second person signs to show the original records were reviewed for accuracy, completeness and compliance. Modifications of a method, and testing of reference standards, reagents and standard solutions, also need complete records.

In practice the most common gaps are calculations done in an unlocked spreadsheet with no record of the formula, chromatograms that were reintegrated without the earlier version kept, and standards prepared without recording the balance used. The FDA's data integrity guidance treats the full audit trail as part of the record for electronic systems, so a printed final report alone is not the complete data. Records are only useful if they are made at the time, so writing results on a glove and copying them later breaks the rule even if every number is right.

How to work it through

1. Record sample identity, source, lot and dates on receipt.
2. Reference the approved method and version and record any permitted deviation.
3. Record weights, volumes, instrument IDs and reagent and standard lots as you work.
4. Keep all raw data and the processing history, including audit trails for electronic systems.
5. Show calculations with units or use a validated, locked calculation tool.
6. Compare the result with the specification, sign and date, then pass the full package to an independent reviewer.

Where the answer stops holding

The regulation lists content, not how to achieve it, so electronic systems need validation and configuration to capture it reliably. A complete record can still be wrong if the underlying measurement is wrong. Review quality depends on the reviewer looking at raw data rather than only the summary.