A near miss is the one where nobody got hurt, and it nearly went otherwise
An unplanned event that had the potential to cause injury or damage and did not. The load that swung and missed. The board that fell into an empty bay. The van that stopped in time. Nothing to record on an injury form, nothing to report to anybody, and by lunchtime nobody remembers it.
HSE describes them as incidents that were narrowly avoided, and as valuable indicators of weaknesses in processes, safeguards or culture. That second half is the reason anybody bothers: a near miss is the only kind of safety information you get beforesomebody is hurt, which makes it the cheapest you will ever have.
The distinction that catches everybody
Near miss and dangerous occurrence get used as though they mean the same thing. They do not, and the difference is not about severity. It is about who decides.
A near miss is whatever your organisation decides to call one. There is no statutory definition, no form and nobody to send it to. You record it because you want to know.
A dangerous occurrence is one of a specific set of events listed in Schedule 2 of RIDDOR: the collapse of a scaffold above a certain height, the failure of lifting equipment, an unintentional explosion, a plant or building collapse, and so on. HSE describes them as unintended, specified events which may cause no reportable injury but which have the potential to cause significant harm. If one happens, you report it, whether or not anybody was anywhere near it.
So the practical question is never "was that bad enough to report". It is "is that on the list".
Why the small ones matter more than the frightening one
The instinct is to record the near miss that made everybody go quiet, and to let the ordinary ones go. That is backwards, and it is backwards for a reason that has nothing to do with conscientiousness.
A single dramatic near miss tells you about one moment. What actually changes a site is the pattern: three unremarkable events in the same corner, on the same shift, around the same delivery. None of the three is worth a conversation on its own. Together they are a finding, and they are invisible unless somebody wrote down all three and later read them as a set.
The thing that kills a near miss system
Blame. It is not close, and every organisation that has lost one lost it the same way: somebody reported a near miss, somebody got a talking-to, and within a month the log went quiet. The site then looked safer than it had ever been, which is the exact opposite of what the numbers meant.
A near miss log that goes to zero has almost never recorded a safer site. It has recorded people deciding not to tell you. That makes the two useful measures of a reporting scheme how many reports you get and how quickly people see something change because of one, in that order.
Where it fits with everything else
A near miss log sits alongside your injury records rather than inside them. The accident record is the statutory account of somebody actually being hurt, and it holds health information about a named person, which is a good reason to keep the two apart.
The other half is what you do next. If a near miss shows a control was missing rather than ignored, that belongs back in the risk assessment, and a near miss is one of the few things that genuinely invalidates one. If it shows people did not know, it belongs in the next toolbox talk, which is what that meeting is actually for.
Common questions
Do I have to report every near miss?
No, and reporting every one to the regulator would be both impossible and unhelpful. What you must report is the specific list of dangerous occurrences in Schedule 2 of RIDDOR, which is a closed set of named events. Everything else is a matter for your own records, and recording those internally is where the value is anyway, because your own near misses are about your own work in a way a national list can never be.
What is the difference between a near miss and a dangerous occurrence?
Who decides. A near miss is whatever your organisation defines it as, recorded because you want to learn from it. A dangerous occurrence is one of a specific list of events written into RIDDOR, with a legal duty to report attached, whether or not anybody was hurt. Every dangerous occurrence would also be described as a near miss in ordinary speech, which is exactly why the two get muddled.
Is a near miss the same as an accident with no injury?
Close, and the useful line is whether anything actually happened. A scaffold board falls and hits nobody: something occurred, it just missed. A worker notices a board is loose before it falls: nothing occurred at all. Both are worth recording and many organisations record them together, but the second is often called an unsafe condition or a safety observation, and separating them makes the pattern easier to read later.
Who should report a near miss internally?
Whoever saw it, and the whole system depends on that being easy and blameless. The single commonest reason a near miss reporting scheme dies is that reporting one gets somebody in trouble, at which point the reports stop and the site looks safer than it was. A form that takes two minutes and a supervisor who thanks people are worth more than any amount of policy.
Does a near miss go in the accident book?
Not usually, and mixing them makes both harder to use. The accident book is the statutory record of injuries to people, and it holds personal and health information about named individuals. Near misses are normally logged separately, which keeps the sensitive record clean and lets the near miss log be shared and discussed openly, which is the point of collecting them.
What do you actually do with them once recorded?
Look for the repeat rather than the dramatic one. A single spectacular near miss gets attention on its own. Three unremarkable ones in the same place, on the same shift, or around the same activity are the finding, and they only become visible if somebody reads the log as a set rather than as a series of incidents. That is the entire argument for recording the small ones.