RPN (risk priority number) multiplies severity × occurrence × detection, each rated 1–10, into a score from 1 to 1,000. Action Priority (AP), introduced in the 2019 AIAG & VDA FMEA Handbook, replaces that multiplication with a lookup table that rates each combination High, Medium or Low, weighing severity first, then occurrence, then detection. The practical difference: RPN treats the three ratings as equally important, so a severe failure can score below a trivial one. AP does not let a low detection or occurrence number hide a high severity.
| RPN | Action Priority (AP) | |
|---|---|---|
| How it is calculated | Severity × occurrence × detection | Lookup table of every S, O and D combination |
| Output | A number from 1 to 1,000 | High, Medium or Low |
| Weighting | All three ratings count equally | Severity first, then occurrence, then detection |
| What drives action | Often a threshold, such as “act above 100” | High: action or documented justification required. Medium: action should be taken. Low: action may be taken |
| Ranking within a level | Implied by the number, though the number is not a true scale | None. AP is not meant to sort failure modes inside a band |
| Source | AIAG FMEA 4th edition and earlier editions of SAE J1739 | AIAG & VDA FMEA Handbook (2019) |
| Main weakness | Equal scores for very different risks; encourages gaming detection | Coarse bands; teams must still decide which High item goes first |
The ratings are ordinal. A severity of 8 is worse than a 4, but it is not “twice as bad”. Multiplying ordinal numbers produces a figure that looks precise but does not behave like a measurement. Three problems follow.
First, very different risks get the same score. 24 different combinations of S, O and D produce an RPN of exactly 120, from a severity 10 safety issue to a severity 2 cosmetic blemish. Second, the scale has holes. Of the 1,000 possible products, only 120 distinct RPN values can occur, so the gaps between them mean nothing. Third, the easiest way to cut an RPN is often to add an inspection and lower detection, which leaves the failure just as likely to happen.
The AIAG 4th edition manual already warned against RPN thresholds for these reasons. The FMEA guide covers the rating scales themselves.
AP keeps the same three 1–10 ratings. Instead of multiplying them, the team finds the row for its severity band, then its occurrence band, then its detection band, and reads the priority. Severity dominates: the same occurrence and detection ratings produce a higher priority for a severe effect than for a minor one, so a good detection rating cannot bury a safety-related failure the way it can in an RPN.
AP labels describe what the team owes. High means the team must identify an action to improve prevention or detection, or record why current controls are adequate, and management review of that rationale is expected. Medium means action should be taken, or the team documents why not. Low means action is optional. The table itself is in the licensed handbook. Use the published version rather than a copy from the internet, since unofficial copies with errors circulate.
You do not need to start over. Keep the structure, functions, failure chains and ratings, check the rating scales against the AIAG-VDA tables (some descriptions changed, particularly for occurrence and detection), then replace the RPN column with AP. Expect the priority list to change: high-severity rows with low RPNs move up, and low-severity rows with high RPNs move down.
Some customers still ask for RPN in their customer-specific requirements. In that case you can show both columns, but act on the one your customer specifies. A PFMEA worksheet with both columns makes the transition easy to show an auditor, and actions should flow into the control plan. The PPAP guide covers how the FMEA fits in the submission.
AP only fixes the arithmetic. If occurrence and detection are guesses, the priority is still a guess. Support occurrence with process data such as defect rates, field returns and capability; see Cpk vs Ppk and the process capability calculator. Support detection with measurement evidence such as a gage R&R study, because a check that cannot reliably separate good from bad parts does not earn a low detection number.
The same AP logic applies to design and process FMEAs; PFMEA vs DFMEA explains what changes between the two.
Example: Two fictional failure modes with the same RPN.
Result: Same RPN, very different risk. RPN cannot tell them apart; AP’s severity-first lookup is built to.
Example: Cutting RPN by adding inspection (fictional). A burr on a machined port: S 7, O 6, D 7.
Result: The inspection lowers the score but the burr is still made at the same rate, and visual checks miss some of them. Prevention cuts the number in half and removes the defect at its source. AP’s weighting points teams toward the prevention action.
Yes, unless your customer requires AP. RPN comes from the AIAG 4th edition and earlier editions of SAE J1739, and some customers still ask for it. Many companies show both during the transition.
Not always. The team must either define an action or document why the current controls are adequate. That rationale should be reviewed and approved.
No. AP depends on the individual S, O and D values, not their product. Two rows with the same RPN can land in different AP bands.
Nothing replaced it directly, but AP’s severity-first table covers the same idea. Some teams still flag severity 9–10 rows as special regardless of AP.
COPLAIN AI PLATFORM
Upload your existing procedures and Coplain AI drafts audit-ready work instructions, job aids and forms from them.
Generate yours automatically with Coplain →Related Tools & Templates