Logo featuring a blue laboratory flask and the text "L@B" in a modern design.
Logo for Advanced Enterprise Agility, emphasizing compliance training.
"L-EAF logo with a graduation cap, symbolizing compliance training."

What do FDA warning letters actually cite? A full read of 2,549 letters

Logo featuring a blue laboratory flask and the text "L@B" in a modern design.
Logo for Advanced Enterprise Agility, emphasizing compliance training.
"L-EAF logo with a graduation cap, symbolizing compliance training."

What do FDA warning letters actually cite? A full read of 2,549 letters

Logo featuring a blue laboratory flask and the text "L@B" in a modern design.
Logo for Advanced Enterprise Agility, emphasizing compliance training.
"L-EAF logo with a graduation cap, symbolizing compliance training."

What do FDA warning letters actually cite? A full read of 2,549 letters

Go back

Search all blogs...

Summarize this article with:ChatGPT logoChatGPTMistral logoMistralClaude logoClaudePerplexity logoPerplexityGrok logoGrok

Author

Published

Reading time

11 min

Author

Published

Reading time

11 min

About 400 companies answered an FDA warning letter by promising more training. In nearly 9 in 10 of those letters, the FDA never cited training. We read all 2,549 non-tobacco letters to see what it did cite.

Last updated: October 2026

Contents

  1. Key Takeaways

  2. What Do FDA Warning Letters Actually Cite?

  3. What Do FDA Manufacturing Warning Letters Cite Most Often?

  4. Why Does Training Appear More Often in Responses Than Violations?

  5. Should You Fix Training or the Procedure First?

  6. How Should QA Teams Review a Warning Letter Finding?

  7. What Are the Limits of This Warning Letter Analysis?

  8. What Should Pharma and Device Manufacturers Do Next?

  9. Frequently Asked Questions

Key Takeaways

  • Procedures lead the findings in manufacturing letters. Of 848 manufacturing letters, 69% cited missing or inadequate procedures, 44% cited procedures not followed, and 6% cited training.

  • Responses often added training. About 400 company responses included training as a corrective action.

  • Training was usually not the written violation. In nearly 9 in 10 of those responses, the FDA had not cited training as a violation.

  • Training can distract from the root cause. Retraining does not repair a procedure that is missing, incorrect, or obsolete.

  • Review the system in order. Check whether the procedure is right, current, and followed before assigning retraining.

About 400 companies answered an FDA warning letter by promising more training. In nearly 9 in 10 of those letters, the FDA had never cited training. We know because we read all 2,549 non-tobacco warning letters the FDA published between January 2021 and October 2026.

The analysis found that about 400 company responses included training, while nearly 9 in 10 of those responses came from letters where the FDA had not cited training as a violation. The practical question is not whether training belongs in a response. It is whether training addresses the finding the inspector actually documented.

What Do FDA Warning Letters Actually Cite?

In the manufacturing letters we read, FDA warning letters most often cite missing or inadequate procedures and failures to follow procedures, not training deficiencies. The figures are in the next section.

We report percentages for the 848 manufacturing letters only. The full set of 2,549 letters also includes letters on other subjects, so we do not pool them.

The pattern matters because a corrective action can look complete while leaving the cited condition intact. A training assignment might produce completion records, but those records do not show that the procedure was adequate or that the process works as written.

The FDA describes warning letters as a principal means of seeking prompt voluntary compliance and establishing prior notice. Its searchable database allows users to filter letters by company, date, office, subject, and response status.[3]

What Do FDA Manufacturing Warning Letters Cite Most Often?

Among 848 manufacturing letters, missing or inadequate procedures were the leading issue by a wide margin.

In the manufacturing subset, 69% of letters cited missing or inadequate procedures. Procedures not followed appeared in 44%. Training appeared in 6%.

Manufacturing issue

Share of 848 manufacturing letters

What it suggests investigators examined

Missing or inadequate procedures

69%

Whether the quality system defined the work adequately

Procedures not followed

44%

Whether personnel and operations followed approved controls

Training cited

6%

Whether training itself was identified as deficient

These percentages should not be added together. One letter can contain more than one issue. A site can have an incomplete procedure, inconsistent execution, and a training weakness in the same inspection record.

The manufacturing results also show why a simple “retrain everyone” response can be too narrow. Counts show how often an issue was cited, not how serious it was. Still, with 69% of manufacturing letters citing missing or inadequate procedures, it is reasonable to check document quality and process design first.

For teams building an evidence trail, auditable SOP training records remain useful after the procedure has been confirmed. The record should connect the learner, the assigned content, and the approved source version.

Why Does Training Appear More Often in Responses Than Violations?

Companies often include training in corrective responses because it is a familiar remediation action, even when the FDA cited a procedure or execution problem instead.

About 400 letters in the review included training in the company’s response. In nearly 9 in 10 of those cases, the FDA had not cited training as a violation. That does not prove the training was unnecessary. It shows that response actions frequently went beyond the wording of the cited finding.

There are reasonable reasons to add training. A revised procedure can change responsibilities. A process failure can reveal that employees misunderstood an instruction. A site may need to show that affected personnel received the updated direction.

The risk comes when training becomes the default explanation. Training cannot make an incomplete procedure complete. It cannot correct a conflicting instruction. It cannot prove that an operator followed a procedure that does not reflect the process in practice.

Should You Fix Training or the Procedure First?

Check the procedure first, then confirm execution, and assign training when the evidence shows that people need instruction.

A practical sequence has three questions:

  1. Is the procedure adequate? Confirm that it describes the required task, controls the relevant risks, and matches the process approved by the quality system.

  2. Is the procedure current? Check the effective version, superseded copies, related forms, work instructions, and change records.

  3. Is the procedure followed? Compare the written steps with records, observations, deviations, batch documentation, and actual operator behavior.

Only after these checks should the team decide whether training is a primary action, a supporting action, or unnecessary. This order prevents a site from measuring training completion while leaving the underlying procedural failure open.

The review does not show whether training updates lagged behind SOP changes. It would be incorrect to claim that the letters prove such a lag. The letters also do not prove the opposite. They simply show that training was often proposed in responses even when it was not the cited violation.

Teams that need a narrower response can use delta training for affected procedure changes. The key control is not the label “delta.” It is the documented link between the approved change and the people who need instruction.

How Should QA Teams Review a Warning Letter Finding?

QA teams should map each FDA observation to the exact system failure before selecting a corrective action.

A useful review starts with the language of the letter, then moves through the controlled documents and operating evidence. The goal is to separate four different conditions that can look similar from a distance.

Condition

Primary response

When training belongs

Procedure is missing

Create and approve the required procedure

After approval, if affected roles need instruction

Procedure is inadequate

Revise the technical and quality content

After the revision changes work expectations

Procedure is current but not followed

Investigate execution, supervision, controls, and behavior

When evidence identifies a knowledge or skill gap

Training is specifically deficient

Correct curriculum, qualification, assignment, or records

As a direct corrective action

This approach also helps L&D avoid owning a quality problem it cannot solve alone. L&D can deliver and document instruction. QA must determine whether the controlled procedure is suitable and whether the process remains compliant.

For a fuller operating workflow, teams can pair this review with a change-control-to-retraining process. The decision should sit inside the change record rather than depend on an informal message to the training team.

What Are the Limits of This Warning Letter Analysis?

The findings are directional, not a complete measure of every FDA inspection issue or every manufacturing failure.

The source set contains 2,549 non-tobacco warning letters published by the FDA between January 2021 and October 2026. It includes only cases the FDA published as warning letters. It does not represent every inspection, observation, untitled letter, settlement, or unpublished compliance discussion.

One AI reviewer assessed each letter. About 11% of letters were unclear and counted as not flagged. That makes the counts conservative, but it also means the categories should not be treated as a perfect human-coded database.

The categories can overlap. A single letter might mention a procedure problem and training in the same context. The percentages therefore describe the share of letters in which an issue was identified, not the share of all observations or the probability that a specific facility will receive the same finding.

The FDA also notes that later interactions with a recipient can change the regulatory status of issues discussed in a warning letter.[3] A published letter should be read as a point-in-time enforcement communication, not as a complete record of the final resolution.

What Should Pharma and Medical Device Manufacturers Do Next?

Manufacturers should make “procedure first” the default triage rule, then add targeted training when the evidence supports it.

For a new warning letter, the response team can use this five-step review:

  1. Extract the cited condition. Copy the specific failure into the response tracker without rewriting it as a training issue.

  2. Identify the controlling document. Record the procedure, work instruction, form, specification, or policy connected to the finding.

  3. Check current status. Confirm the approved revision, effective date, linked documents, and change history.

  4. Test actual execution. Review records and observe the work to determine whether the process matches the document.

  5. Assign remediation by cause. Use document correction, process correction, qualification, supervision, investigation, or training as the evidence requires.

This process keeps a response proportional. It also creates a clearer explanation for the FDA: the company identified the cited failure, corrected the system condition, evaluated affected work, and trained people where training was needed.

L&D leaders should resist accepting every CAPA training request without a source review. If the source is missing, inadequate, or not current, the course team cannot produce a reliable compliance intervention from it.

QA leaders should also avoid the opposite mistake. A procedure can be correct while employees still lack the knowledge or skill to use it. The warning letter analysis does not make training unimportant. It places training in the right diagnostic order.

Skill Studio AI builds training courses from SOPs and updates them when the SOP changes.[1]

Frequently Asked Questions

What do FDA warning letters most often cite in pharma and medical device manufacturing?

In the 848 manufacturing letters we read, missing or inadequate procedures were the most common issue (69%), followed by procedures not followed (44%). Training was cited in 6%.

How often does the FDA cite training in warning letters?

In the 848 manufacturing letters we read, training was cited in 6%. This describes the letters in our analysis and is not a complete count of every training-related FDA observation.

Why do companies include training in FDA warning letter responses?

Training is a familiar corrective action and can help communicate revised expectations to affected employees. The review found about 400 responses that included training, but nearly 9 in 10 of those letters did not cite training as a violation. Training can support remediation, but it should follow a review of the procedure and actual process execution.

Should a company retrain employees after every FDA warning letter?

No. The company should first determine whether the procedure is adequate, current, and followed. Retraining is appropriate when the investigation identifies a knowledge or skill gap, or when a corrected procedure changes employee responsibilities. Retraining alone does not fix a missing, inadequate, or inaccurate procedure.

Can Skill Studio AI help with warning letter remediation training?

Skill Studio AI can turn SOP content into structured training courses and update the training when the source version changes. It can support the delivery and maintenance part of remediation. QA still needs to confirm that the procedure is correct, approved, and suitable before using it as the training source.

What are the main limitations of this 2,549-letter review?

The review covered published non-tobacco warning letters from January 2021 through October 2026, not every FDA inspection or compliance action. One AI reviewer assessed each letter, and 11% were unclear and counted as not flagged. The results are therefore conservative and directional rather than a complete, human-coded regulatory dataset.

References

  1. Skill Studio AI

  2. FDA, Search Databases

  3. FDA, Warning Letters