What is a near miss?
An event that could have hurt someone, and did not. The definition is the easy part — the useful question is what you do with it afterwards.
The short answer
A near miss is an unplanned event that did not result in injury, illness or damage — but realistically could have. The sequence happened. Something interrupted it before anyone was hurt: the operator looked up, the scaffold board held, the forklift was three feet further away than it might have been.
You will also see it called a close call, a near hit, or in some organizations a "good catch" — a term usually reserved for cases where a person deliberately intervened rather than getting lucky. The label matters less than the shared idea underneath: a real failure occurred, and you got the information without paying for it in an injury.
Near miss or incident? The test that actually works
The distinction people struggle with is not near miss versus hazard — it is near miss versus minor incident. A single question resolves most cases: did the harm occur, at any level?
If a worker was struck by a falling wrench and needed a plaster, that is an injury incident. Trivial, but real. If the same wrench landed a metre away and nobody was touched, that is a near miss. The energy was released either way; only the contact differs. Property damage counts as harm too — a dropped load that destroys a pallet of stock is an incident, not a near miss, even though nobody was hurt.
Where organizations genuinely differ is on whether a near miss must involve an actual sequence of events, or whether spotting an unsafe condition also counts. Most programs treat a condition as a hazard report instead. That is a defensible line, and it is worth writing yours down, because a category that means different things to different reporters produces data you cannot trend.
What they look like in practice
| What happened | Why it is a near miss |
|---|---|
| A scaffold plank slipped as a worker stepped off it | The fall sequence started and stopped. Weight distribution, not the system, prevented it |
| A reversing truck stopped a metre from a pedestrian who had walked behind it | Contact was avoided by the driver's reaction, not by a control |
| An energised panel was opened before lockout was verified, and nobody made contact | The procedural failure is complete. The outcome was chance |
| A load shifted in transit and was found leaning against the trailer wall on opening | Restraint failed. Where it came to rest is the only reason there was no injury |
| A worker noticed a frayed sling before lifting with it | Arguably a hazard rather than a near miss — no event occurred. Many programs still want it reported, under whichever category they have defined |
Why they are worth collecting
Injury data tells you what already went wrong. It arrives late, in small numbers, and describes outcomes rather than causes. Near-miss data describes the same failure mechanisms while the cost is still zero, and there is far more of it available if people will report it.
That makes near misses a leading source of information about your program — a measure of what you are catching, rather than what has already hurt someone. Be careful with the stronger version of this argument. You will see fixed ratios quoted, claiming a specific number of near misses precedes every serious injury. Those ratios come from particular studies of particular industries decades ago, and treating them as a law of nature is not sound. The defensible claim is simpler: near misses reveal failure paths cheaply, and a program that collects them has more to work with than one that does not.
There is a counter-intuitive consequence worth understanding. A rising near-miss count is usually good news. It generally means reporting culture is improving, not that the workplace is deteriorating. If your near-miss reports fall to zero, the most likely explanation is that people have stopped telling you.
What a useful report contains
The failure mode of near-miss programs is not that people refuse to report. It is that reports arrive too thin to act on — "almost had an accident at the loading dock" tells you nothing you can fix. A report worth filing captures:
- What happened, in sequence. Not a category, a description. What was being done, what went wrong, and what stopped it.
- Where and when. Specific enough to find the location and correlate it with shift patterns or weather.
- Potential severity, not actual severity. Actual severity is always "none" — that is what makes it a near miss. What matters is how bad the realistic worst case was, and the person who was there is best placed to judge it.
- A photo, where one helps. A picture of the failed restraint or the missing guard removes an entire round of clarifying questions.
- Any immediate action taken. If someone barriered the area or removed the equipment from service, the next person needs to know.
Keep it short. Every additional required field reduces the number of reports you receive, and a program's value comes from volume as much as depth. Detail can be added during review; a report that never gets filed cannot be enriched later.
What to do after one
A report that goes nowhere teaches the reporter not to bother next time. At minimum, a near miss should be reviewed by someone with authority to act, and should end in one of three honest outcomes:
- A corrective action with a named owner and a due date, where something needs to change.
- A documented decision not to act, with the reasoning. This is legitimate and underused — not every report warrants a control change, and saying so openly is better than silence.
- Escalation, where the investigation reveals the potential severity was higher than first reported.
Close the loop with the person who reported it, whichever outcome applies. Feedback is the single strongest predictor of whether they report the next one.
A note on OSHA
Near misses are not recorded on the OSHA injury and illness logs. Those forms record work-related injuries and illnesses meeting the criteria in 29 CFR 1904.4, and an event where nobody was hurt does not meet them. There is no general OSHA requirement to log near misses on the 300.
That is a practical advantage rather than a reason to skip them. Because near-miss reporting sits outside the recordable count, encouraging more of it cannot inflate your injury rates — which removes the perverse incentive that makes some organizations quietly discourage reporting. OSHA does require that employers have a reasonable procedure for employees to report work-related injuries and illnesses, and must not discourage reporting (1904.35); a near-miss program that punishes the reporter tends to undermine that culture generally.