How to Calculate RPN: Severity, Occurrence, and Detection Scoring Guide
RPN (Risk Priority Number) is a numeric score used in FMEA to rank failure modes by risk. It is calculated by multiplying three ratings: Severity (S), Occurrence (O), and Detection (D). The formula is RPN = S × O × D, with each factor scored from 1 to 10. Higher RPN values indicate higher risk and should trigger corrective actions. This guide explains each scale, provides scoring criteria, and walks through a real calculation example.
What Is RPN and When Is It Used?
RPN stands for Risk Priority Number. It is the product of three scores: Severity (how bad the effect is), Occurrence (how often the cause happens), and Detection (how likely you are to catch it before it reaches the customer). The result ranges from 1 to 1,000.
RPN is used in FMEA to prioritize which failure modes deserve attention first. Teams score each failure mode, multiply the three factors, and focus on the highest RPNs. The 2019 AIAG-VDA handbook introduced Action Priority (AP) as an alternative, but many organizations still use RPN for legacy projects or older standards like SAE J1739.
For an overview of FMEA and how scoring fits into the process, see our What Is FMEA guide.
The Severity Scale (1-10)
Severity measures how serious the failure effect is if it reaches the customer or end user. Higher scores mean worse outcomes. Severity is always scored from the customer’s perspective, not from an internal process view.
| Score | Criteria | Example |
|---|---|---|
| 1 | No effect; customer would not notice | Minor cosmetic defect on non-visible surface |
| 2-3 | Very minor; slight inconvenience | Slight noise or vibration within spec |
| 4 | Minor; customer dissatisfied | Reduced performance, noticeable but not critical |
| 5 | Moderate; some loss of function | Partial loss of non-critical function |
| 6 | Significant; major loss of function | Vehicle inoperable but safe |
| 7 | Major; loss of primary function | Complete loss of primary function |
| 8 | Extreme; safety hazard with warning | Brake warning light fails; brakes still work |
| 9 | Serious; safety hazard without warning | Sudden loss of steering or brakes |
| 10 | Critical; regulatory or safety catastrophe | Fire, explosion, or life-threatening injury |
Severity should reflect the worst credible effect. If a bearing failure could cause a shaft to seize and lead to a safety incident, score for that end effect, not just the bearing damage.
The Occurrence Scale (1-10)
Occurrence measures how often the cause of the failure mode is expected to happen. Higher scores mean the cause is more frequent. Occurrence is based on the likelihood of the cause, not the failure mode itself.
| Score | Criteria | Typical Rate (per 1,000 units) |
|---|---|---|
| 1 | Extremely unlikely; failure is theoretical | < 0.01 |
| 2 | Remote; few failures in similar designs | 0.01 – 0.1 |
| 3 | Very low | 0.1 – 0.5 |
| 4 | Low | 0.5 – 1 |
| 5 | Moderate | 1 – 2 |
| 6 | Moderately high | 2 – 5 |
| 7 | High | 5 – 10 |
| 8 | Very high | 10 – 20 |
| 9 | Extremely high; failures are common | 20 – 50 |
| 10 | Almost certain; failure is inevitable | > 50 |
Use field data, warranty data, or process capability when available. If no data exists, use engineering judgment and document the basis. Occurrence should be scored for the cause, not the effect.
The Detection Scale (1-10)
Detection measures how likely you are to find the failure or its cause before it reaches the customer. Unlike Severity and Occurrence, the Detection scale is reversed: lower scores mean better detection, and higher scores mean worse detection.
| Score | Criteria | Example |
|---|---|---|
| 1 | Almost certain to detect | Automated 100% inspection with SPC |
| 2 | Very high chance of detection | Automated inspection; occasional escape |
| 3 | High chance | Multiple manual checks |
| 4 | Moderately high | Single manual inspection |
| 5 | Moderate | Process monitoring; some escape possible |
| 6 | Low | Visual check; easy to miss |
| 7 | Very low | Audit or sampling; many escapes |
| 8 | Remote | No direct check; downstream detection only |
| 9 | Very remote | No detection until customer complaint |
| 10 | No detection method | Failure not detectable until field failure |
Detection is scored for the current controls in place. If you add a new inspection or test, re-score Detection to reflect the improvement. Strong prevention controls (that stop the cause) can also reduce Occurrence.
Step-by-Step RPN Calculation Example
Consider a brake pad wear failure mode in an automotive brake system. Here is how to score and calculate RPN.
Step 1: Define the failure mode and effect.
Failure mode: Brake pads wear beyond safe thickness. Effect: Reduced braking performance; in extreme cases, loss of braking and potential safety hazard.
Step 2: Score Severity.
Loss of braking can cause injury or death. This is a safety hazard without warning. Severity = 9.
Step 3: Identify the cause and score Occurrence.
Cause: Driver ignores brake pad wear indicators and continues driving. Based on warranty data, this happens in about 2 per 1,000 vehicles per year. Occurrence = 5 (moderate).
Step 4: List current controls and score Detection.
Current control: Brake pad wear sensor triggers dashboard warning. Most drivers respond, but some ignore it. Detection = 4 (moderately high; warning exists but not 100% effective).
Step 5: Calculate RPN.
RPN = S × O × D = 9 × 5 × 4 = 180
Step 6: Interpret the result.
An RPN of 180 is high. Most organizations would require action. Possible improvements: improve the wear indicator (loudness, visibility), add a secondary warning, or design for longer pad life to extend the window for replacement.
RPN Thresholds: When to Take Action
There is no universal RPN threshold. Organizations set their own limits based on risk tolerance and industry. Common practices include:
| Threshold | Typical Action |
|---|---|
| RPN > 100 | Mandatory action; must reduce risk before release |
| RPN 80-100 | Action required; document rationale if no action taken |
| RPN 50-80 | Review and consider action |
| RPN < 50 | Monitor; low priority |
Some companies use a Severity cutoff: any Severity of 9 or 10 requires action regardless of RPN. Others use a combination: High RPN or High Severity triggers action. Document your threshold in your FMEA procedure and apply it consistently.
Limitations of RPN and the Move to Action Priority
RPN has several drawbacks that led the automotive industry to adopt Action Priority (AP) in the 2019 AIAG-VDA handbook.
Different combinations, same RPN. The scores 10×1×1 and 5×2×2 both equal 20, but the risk profiles differ. A Severity 10 item may be more critical than a Severity 5 item, yet RPN treats all three factors equally.
Severity is underweighted. Safety-critical failures (high Severity) should be prioritized even when Occurrence or Detection are low. RPN does not enforce this.
Gaps and sensitivity. Many scores cluster in the middle range (80-150), making priorities hard to distinguish. A one-point shift in any factor can move an item above or below a threshold.
Inconsistent calibration. Teams often score differently. Without calibration, RPN comparisons across teams or projects are unreliable.
For these reasons, the AIAG-VDA 2019 FMEA handbook replaced RPN with Action Priority (AP) lookup tables. AP assigns High, Medium, or Low priority in a way that better reflects the importance of Severity.
RPN vs Action Priority (AP)
| Aspect | RPN | Action Priority (AP) |
|---|---|---|
| Calculation | S × O × D (1-1000) | Lookup table; S, O, D → High/Medium/Low |
| Output | Numeric score | Priority level |
| Severity weighting | Equal weight to S, O, D | High S drives High AP even with low O, D |
| Standard | SAE J1739, legacy AIAG | AIAG-VDA 2019 |
| Use case | Legacy projects, some industries | New automotive projects |
Many organizations use both. RPN remains common for existing FMEAs and non-automotive applications. AP is required for new AIAG-VDA projects. Tacit AI supports both methods.
How Tacit AI Approaches This
Tacit AI helps teams score and prioritize failure modes consistently, whether using RPN or Action Priority.
Consistent scoring. Our platform generates FMEA drafts from work orders and manuals, with columns for S, O, D, and RPN (or AP). Teams score against standardized criteria instead of guessing.
Standards-aligned scoring. Our platform generates FMEA drafts from work orders and manuals, with columns for S, O, D, and RPN (or AP). We map to AIAG-VDA, IEC 60812, SAE J1739, and MIL-STD-1629A. Export to Excel or integrate with your existing FMEA software.
Both RPN and AP. We support RPN for legacy projects and Action Priority for AIAG-VDA. Our Dynamic FMEA capability produces FMEAs in the format your organization needs.
Frequently Asked Questions
What is the RPN formula?
RPN = Severity × Occurrence × Detection. Each factor is scored from 1 to 10, so RPN ranges from 1 to 1,000.
What RPN score requires action?
There is no universal threshold. Many organizations require action when RPN exceeds 100. Others use 80 or 120. Some add a Severity cutoff: any Severity 9 or 10 requires action regardless of RPN. Define your threshold in your FMEA procedure.
Why is Detection scored in reverse?
Detection measures how well you can find the failure. A score of 1 means you almost always detect it (good). A score of 10 means you have no detection method (bad). The scale is reversed so that higher Detection contributes to higher RPN, reflecting higher risk.
Is RPN still used in FMEA?
Yes. RPN is still used in legacy projects, SAE J1739, and many non-automotive industries. The 2019 AIAG-VDA handbook introduced Action Priority (AP) as the preferred method for new automotive FMEAs. Many organizations use both depending on the project.
Next Steps
RPN is a simple way to rank failure modes by risk. Use the Severity, Occurrence, and Detection tables in this guide to score consistently. Remember that RPN has limitations; for new automotive projects, consider Action Priority.
Ready to generate FMEAs from your work orders and manuals? Explore Tacit AI’s Dynamic FMEA capability.