A near miss is an event that could have caused injury, illness, or damage but did not. It is defined by what could have happened rather than by its outcome. If the same event with slightly different timing would have hurt someone, it is a near miss.

Are near misses OSHA-recordable?

A near miss with no work-related injury or illness generally is not entered on the OSHA 300 Log under federal 29 CFR Part 1904. There is no general federal Part 1904 requirement for a separate near-miss log. Site, state-plan, or sector-specific requirements may differ. OSHA injury and illness reporting procedures must be reasonable and non-retaliatory.

What is the difference between a near miss and a hazard?

A hazard is a condition that exists, such as a damaged ladder or a missing guard. A near miss is an event that occurred without causing harm. Both belong in a reporting system, and they behave differently: hazards persist until fixed, while the details of a near miss fade within hours.

What should a near miss report include?

What happened, what could realistically have happened, what prevented harm, the conditions at the time such as shift and staffing and running mode, and any immediate action taken. The potential-outcome and what-prevented-harm fields are the ones most often left blank and the ones that make the report useful later.

Why do employees not report near misses?

Usually because nothing visible happened to earlier reports, because reporting takes more effort than the event seemed to warrant, because past investigations produced blame, because a bonus or competition is tied to incident counts, or because the threshold for what counts was never made concrete with examples.

Do safety incentive programs reduce near miss reporting?

They can. OSHA’s guidance permits rate-based incentive programs but warns that withholding a reward because an incident was reported can discourage reporting, and that simply telling employees they will not face retaliation may not be enough to offset a substantial reward tied to low numbers.

Is the 300:29:1 safety pyramid ratio accurate?

It is not a universal conversion rule. Research has found associations between some less-severe and more-severe incidents, but the ratio varies by hazard category, data set, and how severity is defined. Near-miss reporting is valuable because it reveals hazards and control failures, not because 300 reports automatically prevent one fatality.

How do you improve near miss reporting?

Define the threshold with real examples, make reporting possible at the point of work within a minute, acknowledge every report quickly, triage by potential severity rather than actual outcome, investigate the serious and recurring ones properly, communicate closure back to the shift, and remove any metric that rewards low counts.

Near Miss Reporting in Manufacturing: Guide and Template

A near miss is an event that could have caused injury, illness, or damage and did not. The pallet that fell where somebody had been standing a minute earlier. The forklift and the pedestrian who saw each other in time. The valve that was found open before the line was charged.

Nothing happened, which is exactly why the information is so easy to lose. There is no injury to investigate, no downtime to explain, no paperwork that forces the event into a system. Whether anyone ever learns from it depends entirely on whether the person who saw it decided to say something, and on whether saying something has historically been worth the trouble.

This guide covers what a near miss is, what OSHA does and does not require, why plants underreport, a template and checklist you can adapt, and why a site can collect reports for years without getting safer.

What a near miss is, and what it isn't

What defines a near miss is its potential, not its outcome. Shift the timing by a few seconds and the same event is a recordable injury. Luck separated the two, and luck is not a control.

That definition matters because it decides what belongs in the system. If a plant only logs events that came close to hurting someone in an obvious way, it misses the larger category: the conditions that would have caused harm had anyone been in the wrong place. An unguarded opening on an unstaffed platform is a hazard even if nobody walked past it. If a worker steps into that opening but avoids falling, the event is a near miss. Both should be reported through an appropriate safety process.

It is not the same as a hazard observation, though plants often use one form for both. A hazard is a condition that exists. A near miss is an event that occurred. Both are worth capturing, and the distinction is worth keeping because they behave differently. Hazards tend to persist until somebody fixes them. Near misses are moments, and the details evaporate within hours.

It is also not an incident report with a smaller font. An incident investigation starts from a known outcome and works backward. A near miss investigation has no outcome to anchor to, so it has to reason about what could have happened, which is a harder question and the reason these reports are so often closed with a single line.

What OSHA actually requires

Federal OSHA injury and illness recordkeeping rules are not a general requirement to log every near miss. State-plan rules, site procedures, and sector-specific obligations can add requirements. See 29 CFR Part 1904 and OSHA reporting criteria.

This is where a lot of plant safety programs carry a misconception, in both directions.

Near misses are not OSHA-recordable. The recordkeeping rule at 29 CFR 1904 covers work-related injuries and illnesses. An event with no work-related injury or illness generally does not go on the OSHA 300 Log, and there is no federal requirement to maintain a near miss log at all.

What OSHA does require is a reporting system that works. Under 1904.35(b)(1)(i), employers must establish a reasonable procedure for employees to report work-related injuries and illnesses promptly and accurately, and a procedure is not reasonable if it would deter or discourage a reasonable employee from reporting. Section 11(c) of the OSH Act already prohibited retaliation for reporting, and the rule wrote that prohibition explicitly into 1904.35(b)(1)(iv).

The part worth reading closely is OSHA’s 2018 guidance on safety incentive programs. Rate-based incentive programs stay permissible as long as they are not run in a way that discourages reporting, but if an employer withholds a prize or a bonus because an injury was reported, OSHA would decline to cite only where the employer had taken adequate precautions to ensure employees feel free to report. The guidance goes further: telling employees they are encouraged to report and will not face retaliation may not by itself be adequate, particularly when reporting costs them a substantial reward.

A bonus tied to low reported injury counts can create pressure to stay quiet if it is administered in a way that discourages reporting. OSHA permits rate-based incentives when appropriate safeguards prevent that result. See its October 2018 guidance.

The compliance distinction is narrower than “every near miss must go on the OSHA 300 Log.” A useful reporting culture still matters: a procedure that discourages injury and illness reporting is a separate problem from an empty voluntary near-miss log.

Why plants underreport

When a site suspects near-miss underreporting, the first questions should be about reporting friction and what happens after an observation is raised. Workers may care about preventing harm and still have little reason to trust the reporting process.

Nothing visible happens afterward. Someone files a report and hears nothing back. No acknowledgement, no action, no explanation of why the thing they flagged was or was not a problem. Reporting becomes unpaid administrative work with no observable output, and people stop doing unpaid work with no observable output.

The report costs more than the event seemed to. A form in the supervisor’s office, filled in at the end of a twelve-hour shift, competes directly with going home. Reports get made when they are easy to make, which usually means within a minute of the event and at the place it happened.

Reporting has historically produced blame. Where investigations have ended with somebody being disciplined, the workforce has learned what a report is for. This is the hardest of the causes to reverse, because it is a belief built on evidence.

The metric punishes the report. Incident-rate bonuses, safety competitions between shifts, and streak counters all create a reason to stay quiet. A plant that rewards low numbers will reliably get low numbers.

Nobody is sure it counts. When the definition is vague, people resolve the ambiguity by not filing. If the guidance is "report near misses" with no examples, the threshold each person applies is their own.

A low report count alone does not prove a site is safe. It may indicate low exposure, an unusually effective control program, or simply that workers do not see enough value or trust in reporting. Review the reporting process before interpreting the number.

What a near miss report should capture

Most forms collect what happened and stop there. A report that can produce a fix covers five things:

What happened. The event in plain language, including where and when, and what the person was doing at the time.
What could have happened. The realistic worst outcome rather than the dramatic one. Priority gets set from this field, and it is the one most often left blank.
What made the difference. Why it stayed a near miss. If the answer is that someone happened to look up, that is a very different finding from a guard doing its job.
Conditions at the time. Staffing, shift, lighting, time into the shift, whether the line was running normally or in a non-standard mode. These are what make patterns visible later.
Immediate action taken. What was done right then, kept separate from what still needs to be done.

The second and third fields are the ones that separate a useful report from a filed one. Without them, a reviewer months later has an event with no way to judge how close it came or what actually protected the person.

Calculate your near-miss follow-up rate

Track the share of near-miss reports acknowledged and closed, and the number still needing attention. These are process measures, not a prediction of injury reduction.

Near-miss reporting follow-up calculator

Calculate process response rates for the same reporting period.

Near-miss reports in period

Reports acknowledged

Reports closed with a documented decision

Acknowledgement rate

85.0%

Closure rate

62.5%

Still open

Acknowledged and closed counts are capped at total reports. Review closure quality, severity, and time to action separately. Higher reports or closure rates do not directly demonstrate fewer injuries or OSHA compliance.

Reporting methods compared

| --- | --- | --- | --- |

| Paper form | Forms kept at a station or in the supervisor’s office, filled in and handed over | No devices needed; works anywhere in the plant; low barrier for a workforce that distrusts systems | Filled in from memory at end of shift; easy to lose; no aggregation without someone typing them up |

| Verbal to supervisor | Raised in conversation, supervisor decides whether to record it | Fastest possible reporting; captures nuance | Nothing exists unless the supervisor writes it; filtered through one person’s judgment of what counts |

| Digital form on a shared terminal | Kiosk or shared PC on the floor | Persists and aggregates; available to every worker including contractors | Still a walk and a queue; terminals sit in offices more often than at the point of work |

| Mobile or voice capture at the point of work | Reported on a phone or by voice where the event happened | Captured while details are fresh; lowest friction; timestamped and located | Needs coverage in the plant; needs a workforce that trusts what happens to the report |

No format fixes a culture problem. A site where reports produce blame will underreport on the best software available. Format decides how much friction sits on top of whatever willingness already exists.

How to run near miss reporting step by step

Step 1: Define the threshold with examples

Write down what counts, and include five or six real examples from your own plant. Vague definitions get resolved by not filing. Examples set the threshold far more effectively than a definition does.

Step 2: Put reporting where the work is

The report should be possible within a minute of the event, at the place it happened, without leaving the floor or waiting for a terminal. Every additional step costs reports, and it costs them unevenly. The hardest events to report are the ones that happen at the worst moments.

Step 3: Acknowledge every report, fast

Someone confirms receipt to the person who reported, within the shift where possible. This single step does more for reporting rates than any campaign, because it answers the question that stopped people reporting in the first place.

Step 4: Triage on potential, not on outcome

Rank by what could have happened. A near miss with a realistic fatal outcome outranks a minor injury that did occur, and a triage process that sorts by what did happen will systematically deprioritize the most valuable reports in the system.

Step 5: Investigate the ones that warrant it

Not every report needs a full investigation. The ones with serious potential, and the ones that keep recurring, get root cause analysis with the same rigor as an injury. Reasoning about a hypothetical outcome is harder than reasoning backward from a real one, so these need more discipline, not less.

Step 6: Close the loop out loud

Tell the reporter what happened, and tell the shift. Reporting rates are driven by what people have seen happen to previous reports, which means visible closure is not a courtesy. It is the input to next month’s numbers.

Step 7: Look across reports, not just at them

The value is in the pattern. Same area, same shift, same hour into the shift, same task. A single report is an event; forty reports are a map of where the plant is close to hurting someone.

Near miss reporting checklist

Definition and access

Threshold defined in writing, with real examples from this plant
Reporting possible at the point of work, within a minute
Available to contractors and temporary staff, not just employees
Anonymous route available for people who want it

The report itself

Captures potential outcome, not just what happened
Captures what prevented harm
Captures conditions: shift, staffing, hours in, running mode
Immediate action recorded separately from outstanding action

Response

Every report acknowledged to the reporter
Triage ranks by potential severity
Serious-potential and recurring reports get a real investigation
Actions have named owners and due dates
Closure communicated back to the reporter and the shift

Culture and metrics

Incentive programs checked for unintended pressure not to report
Reporting rate treated as a health signal, not a problem to reduce
Reports reviewed for patterns, not only case by case
Patterns compared across shifts, areas, and sites

Near miss report template

A starting point to adapt. Keep the fields fixed so reports stay comparable across shifts and areas.

NEAR MISS REPORT

Reported by (optional):        Date / time of event:

Area / line / asset:           Shift:  Days / Swing / Nights

Hours into shift:              Reported at (time):

1. WHAT HAPPENED

Plain description of the event:

What the person was doing at the time:

2. WHAT COULD HAVE HAPPENED

Realistic worst outcome:

Severity if it had occurred:  Minor / Serious / Fatal

3. WHAT PREVENTED HARM

Control, guard, procedure, or chance:

4. CONDITIONS AT THE TIME

Staffing / lighting / running mode (normal or non-standard):

Anything unusual about the day:

5. IMMEDIATE ACTION

What was done right away:

6. FOLLOW-UP

Action | Owner | Due | Status

7. CLOSURE

Acknowledged to reporter (date):

Outcome communicated to shift (date):

Assess your near-miss reporting process

Answer ten questions about reporting access, response, and learning across shifts. This is a self-assessment, not an OSHA compliance determination.

How much is your plant not hearing?

Answer for your actual practices, not the documented ideal. Yes = 2 points, Partly = 1, No = 0. Your responses stay in this browser page.

0 of 10 answered

A note on the safety pyramid

Almost everything written about near misses cites a ratio. Three hundred near misses per serious injury, or six hundred, depending on the source. These come from Herbert Heinrich’s 1931 work and Frank Bird’s 1960s revision of it, and they are worth understanding before repeating.

Heinrich proposed a 300:29:1 ratio relating near misses and minor injuries to a single major injury, and although the idea has been hugely influential in safety policy, its standing has been questioned. The criticism targets both the fixed ratios themselves and the assumption that preventing low-severity events prevents high-severity ones. Bellamy’s analysis of 23,030 Dutch occupational accidents found that relationships differ by hazard category. Yorio and Moore’s analysis of more than 25,000 establishments found associations between lower-severity incidents and later serious outcomes, but the triangle’s form depended on how severity was defined. Those findings do not support a universal fixed ratio.

Heinrich was also narrower than his popularizers. His own text framed the ratio as an average across a unit group of 330 accidents of the same kind involving the same person, and he noted that the figure was persistently misquoted.

None of that undermines near miss reporting. It undermines one specific claim: that collecting three hundred reports buys you a prevented fatality. The defensible reason to run the program is narrower and holds up better. A near miss can reveal a failure or vulnerability in controls before an injury, and you cannot fix a pattern you cannot see. That argument needs no ratio, which is useful, because a universal fixed ratio is not supported by those studies.

Why reports stop producing learning

There is a failure mode past underreporting, and it catches the sites that have already done the hard work. Reporting rates come up. The forms are good. Reports get acknowledged. And the same near misses keep happening.

What has broken by then is not collection. It is retrieval. Reports go into a system that stores them as records, one row per event, categorized by type. Months later the same thing happens in a different area on a different shift, and nothing surfaces the earlier one. Each report was handled. No one saw the pattern, because seeing it would have required someone to already suspect it was there.

It gets worse across sites. The same event is categorized as a slip at one plant, a housekeeping issue at another, and a guarding problem at a third, and the group roll-up aggregates categories rather than causes. Three instances of one problem look like three unrelated small things.

Then there is what the record does not hold. A reviewer decided this one did not warrant investigation, and the reasoning behind that call is nowhere. Someone ruled out a cause, and the fact that it was ruled out is not written down. Whoever next reads the report cannot tell the difference between a question that was considered and rejected and one nobody thought to ask.

Institutional knowledge leaves a plant this way too, and so does the reasoning at shift handover, where the log takes the status and drops the thinking. The near miss report holds the event. The judgment around it, whether this was serious, what it resembled, what was already tried, tends to live in the head of whoever handled it. When that person moves on, the reports remain and the learning does not.

How SteelTree helps close the loop

SteelTree preserves the operational decision trail around near misses: what was reported, what it was judged to be, what was done, and what happened the last time something like it came up. The next person reading a report sees what the plant already knows instead of starting from one row in a log.

See SteelTree safety compliance use cases for how teams can review recurring patterns and the decisions behind follow-up work.

Near Miss Reporting in Manufacturing: Why Sites Underreport and How to Fix It