AC 120-92D
The FAA 5×5 risk matrix, explained
Likelihood times severity, banded into three decisions. Simple to describe, and routinely applied in ways that will not stand up to an inspector's questions.
The two axes
Every hazard is scored on two independent scales. Multiply them and you get a risk score between 1 and 25.
Likelihood — how often you expect it
| Score | Level | Meaning |
|---|---|---|
| 5 | Almost certain | Expected to occur frequently |
| 4 | Likely | Expected to occur several times |
| 3 | Possible | Could occur sometime |
| 2 | Unlikely | Not expected, but possible |
| 1 | Rare | Extremely unlikely to occur during operations |
Severity — how bad it is if it happens
| Score | Level | Meaning |
|---|---|---|
| 5 | Catastrophic | Fatal injury or significant property damage likely |
| 4 | Major | Serious injury or major damage |
| 3 | Moderate | Injury requiring treatment, or moderate damage |
| 2 | Minor | Minor injury or minor damage |
| 1 | Negligible | No injury or damage to property |
The three bands
| Score | Band | What it obliges you to do |
|---|---|---|
| 15–25 | Unacceptable | Immediate corrective action required. The activity should not continue in its current form. |
| 5–14 | Acceptable with controls | A management decision is required. Controls must be applied and their effect verified. |
| 1–4 | Acceptable | Acceptable as is. Record it and monitor. |
Initial and residual: the part that gets skipped
A single score is not enough. Part 5 asks you to demonstrate that risk controls work, and you cannot demonstrate that with one number. Every hazard needs two:
- Initial risk — scored before you do anything about it. This is the honest picture of the hazard as found.
- Residual risk — scored after your controls have been applied and given time to take effect.
The difference between the two is your evidence of control effectiveness. An operator who records only initial scores has a hazard log. An operator who records both has a safety assurance programme.
Three ways operators get this wrong
Scoring the outcome that happened rather than the credible worst case
A tail strike that caused minor damage is not automatically severity 2. Severity asks what the credible consequence of that hazard is, not what it happened to be on the day. Score the hazard, not the anecdote.
Deflating scores to avoid the paperwork
A 15+ score triggers immediate action, which is inconvenient. The temptation to call something "possible" rather than "likely" is real, and it is visible to anyone who reads your register with a critical eye. It also defeats the entire purpose.
Assessing in isolation
Three separate reports of the same unstable approach into the same airport are not three "possible" events. Together they are evidence of something closer to "likely". Trend tagging exists so the pattern surfaces before you score the third one on its own.
Where the matrix comes from
The 5×5 matrix described here is the standard one defined in FAA Advisory Circular 120-92D for aviation SMS. It is seeded in this platform ready to use, so you are scoring on the same scale your inspector reads. Some operators maintain additional matrices for specific domains such as maintenance or flight operations; the same initial-and-residual discipline applies to all of them.
See whether it fits your operation
Twenty minutes, no card. Tell us what certificate you hold and where your implementation plan stands, and we will show you the parts that apply to you.