TRIR is the number of OSHA recordable cases per 100 full-time workers in a year. The formula is (recordable cases × 200,000) ÷ hours worked. Seven recordable cases over 250,000 hours gives a TRIR of 5.6. Everything hard about the number is in the two inputs, not in the arithmetic.
The formula, in OSHA's own words
OSHA sets it out in a 2016 letter of interpretation:
(Number of injuries and illnesses X 200,000) / Employee hours worked = Incidence rate
OSHA, Clarification on how the formula is used by OSHA to calculate incident rates, August 23, 2016
The same letter explains the constant: 200,000 is the number of hours 100 employees working 40 hours per week, 50 weeks per year would work
. Dividing by your own hours and multiplying by that base converts your year into cases per 100 full-time workers, so a 12-person crew and a 4,000-person plant land on the same scale.
The constant never changes. It does not become 240,000 because your crews work 48 weeks of overtime, and it does not shrink for a part-time workforce — the hours you divide by already handle that.
Where the two numbers come from
Both inputs are already on the OSHA 300A summary you post every February. The case count is the sum of the four case boxes; every recordable case sits in exactly one of them, chosen on its most serious outcome.
| Box | What it counts | In TRIR? |
|---|---|---|
| G | Total number of deaths | Yes |
| H | Cases with days away from work | Yes |
| I | Cases with job transfer or restriction | Yes |
| J | Other recordable cases | Yes — the box people forget |
| K and L | Total days away, and days of transfer or restriction | No — days never enter a case-based rate |
Column J is where most understated rates come from. A case that ended with stitches, a prescription or a course of physical therapy — medical treatment beyond first aid — is recordable even though nobody missed a shift and nobody was put on light duty. It belongs in the count.
Which hours count
Hours actually worked, from payroll. That means:
- Include overtime, and the hours of salaried, hourly, part-time, temporary and seasonal workers — including temporary help service workers under your day-to-day supervision.
- Exclude vacation, sick leave, holidays and any other non-work time, even if the employee was paid for it.
If your system only tracks hours paid, OSHA publishes an optional worksheet with the recordkeeping forms: multiply your full-time employees by the annual work hours of a full-time employee, then add overtime and the hours worked by everyone else. It is an estimate, and OSHA accepts it as one.
A worked example
A specialty contractor with about 125 people on the payroll works 250,000 hours in a calendar year. The 300A shows no fatalities, 2 cases with days away, 1 case with restricted duty and 4 other recordable cases.
TRIR = (7 × 200,000) ÷ 250,000 = 5.6
Read that as 5.6 recordable cases per 100 full-time workers. The 250,000 hours are worth 125 full-time equivalents (250,000 ÷ 2,000), which is a useful cross-check against the headcount you wrote on the summary.
One more number is worth knowing: each case is worth 200,000 ÷ your hours in rate points. Here that is 0.8. Add one slip in a wet December and the year reads 6.4. At 40,000 hours a single case is worth 5.0 points, which is why a small contractor's annual rate says more about luck than about safety. You can run all of this, including the per-case sensitivity, in the TRIR calculator.
Why two people get two answers from the same log
TRIR is not defined in 29 CFR Part 1904 — the standard tells you to keep the Log, total the columns, certify the summary and post it. It never asks for a rate. So the definition comes from whoever is asking, and the variations are real:
- Fatalities in or out. OSHA's VPP application asks applicants for a Total Case Incidence Rate computed on recordable non-fatal injuries and illnesses. Most companies calling their number TRIR include column G. Same log, different rate.
- Contractor hours in or out. A VPP construction applicant includes all site employees, contractors and subcontractors included. A prequalification form usually wants your own employees only.
- Establishment or company. The 300A is kept per establishment. A company-wide rate is a different figure, and comparing one against an establishment-level benchmark is meaningless.
None of these is cheating. Reporting a rate without saying which convention you used is the problem. Give the case count and the hours alongside the rate and the ambiguity disappears.
TRIR, DART and lost time rate
Three rates, one formula, three different case counts. All of them use the same 200,000-hour base, so they are directly comparable and always fall in the same order.
| Rate | Cases counted | Example above |
|---|---|---|
| TRIR | G + H + I + J — every recordable case | 7 cases, rate 5.6 |
| DART | H + I — days away, restricted work or transfer | 3 cases, rate 2.4 |
| Lost time case rate | H — days away only | 2 cases, rate 1.6 |
The one OSHA acts on is DART: the Site-Specific Targeting program builds its inspection list from the DART rates in the Form 300A data establishments submit electronically. That makes columns H and I worth getting exactly right. Work yours out in the DART rate calculator.
What counts as a good TRIR
There is no OSHA threshold that makes a rate acceptable, and anyone quoting one is quoting a customer's target, not a rule. A rate means something only against the published average for your industry and your size.
- Use your NAICS code, at the most precise level available. The Bureau of Labor Statistics publishes incidence rates by industry and by establishment size, and OSHA's VPP program asks applicants to compare against exactly that.
- Compare like sizes. OSHA's 2016 letter makes the point directly: a single case has a much greater effect on a small establishment's rate than on a large one's, which is why BLS breaks rates into five size categories.
- The broad national number is 2.4. OSHA's own statistics page puts injuries and illnesses at 2.4 incidents per 100 workers in 2023, down from 10.9 in 1972. Useful as an order of magnitude; useless as a benchmark for a roofing crew.
- Three years beats one. For anything under about 200,000 hours a year, combine several years of cases and hours — keeping the 200,000 constant — before drawing a conclusion.
Five mistakes that change the answer
- Forgetting column J. Medical treatment beyond first aid with no lost time is still a recordable case.
- Counting days instead of cases. Columns K and L are days. They belong to severity, not to TRIR.
- Double-counting a case that had both days away and later restricted duty. It is one case, recorded in column H, with its days split between K and L.
- Hours paid instead of hours worked. See the sanity check above.
- Managing the rate instead of the risk. A reporting procedure that would deter or discourage a reasonable employee from accurately reporting an injury is not a reasonable procedure under 29 CFR 1904.35(b)(1)(i), and discharging or discriminating against an employee for reporting is prohibited outright by (b)(1)(iv). Rate-based safety incentives sit close to that line.
Bottom line
- TRIR = (recordable cases × 200,000) ÷ hours worked, and 200,000 is 100 full-time workers for a year.
- Cases come from columns G, H, I and J of the 300A. Days never enter the number.
- Hours are hours actually worked. Paid leave is not worked time.
- Always publish the case count and the hours next to the rate — it is the only way anyone can check it.
- Totalling the summary first? Run the boxes through the 300A summary helper and the rates come out with them.