Corrective Action Program: A Clear NQA-1 Guide for 2026

A corrective action program is the part of a nuclear quality assurance system that decides whether problems get smaller or repeat. NQA-1 and 10 CFR 50 Appendix B both require it, and both draw the same line that ordinary quality systems do not: conditions adverse to quality are corrected, while significant conditions adverse to quality also require the cause to be determined and corrective action taken to preclude recurrence.

This guide covers that distinction, what each path obliges you to do, and the three failures that turn a corrective action program into a backlog.

The distinction that drives everything

A condition adverse to quality is any deficiency — a failure, malfunction, defective item, deviation, or nonconformance. The requirement is that it is identified and corrected.

A significant condition adverse to quality carries more. The cause has to be determined, corrective action taken to preclude recurrence, and the condition, its cause and the corrective action reported to appropriate levels of management. The word doing the work is preclude: fixing the item is not enough if the mechanism that produced it remains in place.

Which means the classification step is the highest-leverage decision in the program — and the one most often made by whoever raises the condition, with no criteria in front of them.

Making classification defensible

A corrective action program needs written criteria applied at screening, by a person with the authority and knowledge to judge. Useful triggers for the significant path:

• Any condition affecting nuclear safety, a safety-related item or a critical characteristic.
• Recurrence — the same or a similar condition previously identified. Repetition is itself evidence the earlier action did not preclude anything.
• A programmatic breakdown rather than an isolated error: a procedure that cannot be followed, training that did not happen, a control that was never implemented.
• A condition that escaped to the customer or downstream, including anything shipped.
• A commercial-grade dedication failure, since it calls the basis of acceptance into question — see our guide to commercial grade dedication.

Screening should be a daily rhythm, not a monthly meeting. Conditions age badly: evidence disappears, the people involved forget, and a condition that could have been understood in a week takes a month.

Cause determination that finds a cause

Three habits separate real cause analysis from the appearance of it:

1. Stop at a mechanism you can act on, not at a person. “Operator error” is a description of the event, not a cause. Why was the error possible, likely, or undetected?
2. Match the method to the significance. A simple apparent-cause evaluation for most conditions; a formal root cause analysis for the significant ones, performed by somebody independent of the work.
3. Test the extent of condition. Where else could this exist? A cause that produced one nonconforming weld probably produced others, and the extent-of-condition review is what turns a single correction into a preclusion.

Effectiveness review is the step that gets dropped

An action taken is not an action that worked. Set a review point when the action is defined — a date, a sample, or a trigger — and record the finding, including when the answer is that the condition recurred. A corrective action program whose entries all close as effective is not measuring effectiveness.

Three failures that turn a corrective action program into a backlog

Classification inflation, then paralysis. Everything is marked significant, the root cause queue grows past capacity, and genuinely significant conditions wait behind trivia. The fix is criteria plus an authorized screener, not more analysts.

Low-threshold reporting without low-threshold handling. A strong safety culture encourages reporting minor conditions — which is right — but the program has to have a proportionate path for them, or people learn that raising something creates work and stop.

Trending that nobody uses. The program generates the best data in the organization: what fails, where, how often. If the trend analysis does not change training, procedures or supplier decisions, the data is being collected for its own sake.

Where it sits in the wider system

The corrective action program is the connective tissue of an NQA-1 quality assurance program. Audits and surveillances feed it, the graded approach determines how much rigour each condition attracts — see the graded approach — and the regulatory baseline sits in 10 CFR 50 Appendix B, whose criteria the program has to satisfy alongside NQA-1 itself.

Frequently asked questions

What is a condition adverse to quality?
A deficiency such as a failure, malfunction, deviation, defective item or nonconformance. It must be identified and corrected.

What makes a condition significant?
Your classification criteria, applied consistently — typically safety impact, recurrence, programmatic breakdown or escape to the customer. Significant conditions additionally require cause determination, action to preclude recurrence, and reporting to management.

Is a corrective action program required by 10 CFR 50 Appendix B?
Yes. The corrective action criterion carries the same two-tier structure, which is why NQA-1 programmes and Appendix B programmes use the same vocabulary.

Who should perform root cause analysis?
Somebody independent of the work being analysed, trained in the method. Self-analysis by the responsible group produces causes that stop conveniently short.

How long should conditions stay open?
Long enough to do the analysis properly and no longer. Track age and overdue counts, and treat a growing backlog as a condition in its own right.

Where this leaves you

Build the corrective action program around the classification decision: written criteria, an authorized screener, and daily screening. Reserve formal root cause for genuinely significant conditions, always ask where else the cause could have acted, and set the effectiveness review when the action is defined rather than after it closes. Then use the trend data to change something — a program that records conditions faithfully and changes nothing is an archive, not a corrective action program.

References

• 10 CFR 50 Appendix B — the quality assurance criteria, including corrective action.
• ASME NQA-1 — quality assurance requirements for nuclear facility applications.

More on nuclear quality assurance

• The corrective action program — you are here
• NQA-1 explained
• The NQA-1 graded approach
• 10 CFR 50 Appendix B

Condition reports, cause analysis forms and effectiveness review records are in the Nuclear Quality Assurance Toolkit, or start with the free ISO templates.