What is the formula for TRIR?

TRIR = (number of OSHA-recordable cases × 200,000) ÷ total hours worked by all employees during the year. The case count comes from your OSHA Form 300 log and the hours from payroll, actual hours worked, excluding vacation and leave. The result is recordable injuries per 100 full-time workers.

Why is TRIR based on 200,000 hours?

200,000 hours is the equivalent of 100 employees working 40 hours a week for 50 weeks a year. Using it as a constant normalizes injury counts across companies of different sizes, so a 40-person shop and a 4,000-person plant can be compared on the same per-100-workers scale. The convention is published by the Bureau of Labor Statistics.

What is a good TRIR for manufacturing?

Compare against your specific NAICS code in the BLS industry tables rather than a generic number. Private industry overall ran 2.3 per 100 full-time workers in 2024, and manufacturing has run somewhat above that average, with wide variation by subsector. Beating your own subsector's rate is a defensible claim; beating a poster number is not.

What is the difference between TRIR and DART?

Same formula, different numerator. TRIR counts all recordable cases; DART counts only cases involving days away from work, restricted work activity, or job transfer, the more severe subset. DART is always less than or equal to TRIR, and the two together say whether your injuries are frequent, severe, or both.

Is a low TRIR proof that a plant is safe?

No. TRIR is a lagging indicator, it counts past injuries, not present risk, and it says nothing about near misses, bypassed guards, or skipped procedures. It's also noisy at small sites and improvable by underreporting. A low TRIR alongside healthy leading indicators is evidence of safety; a low TRIR alone is just a low number.

How can one injury give a small company a high TRIR?

Because the denominator is small. A 40-person plant works about 80,000 hours a year, so a single recordable produces a TRIR of 2.5, already above the 2024 national private-industry average of 2.3. Small sites should trend TRIR over rolling multi-year windows or use raw case counts instead of reacting to single-year swings.

Do near misses count in TRIR?

No. TRIR counts only OSHA-recordable injuries and illnesses, so near misses never appear in it, which is precisely its blind spot. Tracking near misses separately gives you the early-warning signal TRIR can't, and a healthy near-miss-to-recordable ratio is one of the harder metrics to game.

How to Calculate TRIR (and What It Hides)

The 200,000-hour formula, a worked example, DART, and the three ways the most-quoted safety number misleads.

TRIR (Total Recordable Incident Rate) is the number of OSHA-recordable injuries and illnesses per 100 full-time workers per year. The formula: recordable cases multiplied by 200,000, divided by total hours worked. The 200,000 is 100 workers times 40 hours times 50 weeks, a normalizing constant, nothing more.

TRIR is the most-quoted safety number in industry: customers ask for it, insurers price on it, corporate dashboards rank plants by it. It's also routinely misread. This guide covers the math, a worked example, the DART variant, where to find honest benchmarks, and the three ways TRIR misleads, especially in smaller plants.

What is the TRIR formula?

TRIR = (number of recordable cases × 200,000) ÷ total hours worked by all employees. Both inputs come straight from your OSHA paperwork: the case count is your Form 300 log and the hours-worked figure is the same one that goes on your 300A summary. The formula is published by the Bureau of Labor Statistics (How to Compute Your Firm's Incidence Rate) and used the same way by OSHA.

Hours worked means actual hours, no vacation, sick leave, or holidays. Include temporary and part-time workers you supervise day to day: their cases go on your log, so their hours belong in your denominator.

Both inputs come from your OSHA recordkeeping: the 300 log supplies the cases, the 300A hours field supplies the denominator.

How do you calculate TRIR? A worked example

Take a plant with about 150 employees across two shifts.

Count recordable cases for the year. The Form 300 log shows 4: two lacerations requiring sutures, one shoulder strain with restricted duty, one hand fracture with days away. (First-aid-only cases never enter the count, the line is drawn in 29 CFR 1904.)
Total the hours actually worked. Payroll shows 310,000 hours for the year, overtime included, leave excluded.
Multiply cases by 200,000: 4 × 200,000 = 800,000.
Divide by hours worked: 800,000 ÷ 310,000 = 2.58.
Read it as a sentence: if 100 people worked full-time at this plant for a year, you'd expect about 2.6 recordable injuries at the current rate.

What is DART, and how is it different?

DART (Days Away, Restricted, or Transferred) uses the identical formula but counts only the subset of recordables severe enough to involve days away from work, restricted work, or a job transfer. In the example above, 2 of the 4 cases qualify (the strain and the fracture): DART = 2 × 200,000 ÷ 310,000 = 1.29. DART is always ≤ TRIR, and the gap between them says something: a TRIR full of DART cases means your injuries are serious, not just frequent. Insurers and customer prequalification forms typically ask for both.

What's a good TRIR? Benchmarks and context

"Good" only means anything against your own industry, because baseline hazard varies enormously. The primary sources:

Private industry overall recorded a TRIR of 2.3 in 2024 down from 2.4 in 2023 and the lowest in the series, about 2.5 million recordable cases in total (BLS, Employer-Reported Workplace Injuries and Illnesses).
Manufacturing has run above the all-industry average in recent years, around half a point higher, with wide spread between subsectors; look up your own NAICS code in the BLS industry rate tables rather than borrowing a generic number.
The formula and the 200,000-hour basis are documented at BLS.

A plant beating its NAICS average has a defensible claim. A plant comparing itself to "industry average 3.0" it saw on a poster does not.

What does TRIR hide?

Three things, and they compound.

It's a lagging indicator

TRIR counts injuries that already happened. It says nothing about the near miss last Tuesday, the bypassed interlock, or the lockout/tagout step everyone skips on Fridays. A plant can run a beautiful TRIR right up until the day it doesn't, the precursors were never measured. Leading indicators, near-miss reports filed, job safety analyses completed and revised, toolbox talks held, corrective actions closed on time, measure the work that prevents the injury, not the injury.

TRIR belongs on the dashboard, next to the leading indicators that actually predict it.

Small numbers make it noisy

The formula divides by hours, and small plants don't have many. A 40-person plant works roughly 80,000 hours a year. Zero recordables: TRIR 0.0. One recordable, a single stitched finger, and TRIR jumps to 2.5, worse than the national all-industry average. Two, and it's 5.0, double the average. The difference between "best in class" and "twice the national rate" is literally one incident, which makes year-over-year TRIR moves at small sites mostly statistical noise. Trend it over rolling multi-year windows, or use case counts directly, before drawing conclusions.

At 80,000 hours a year, every single recordable moves TRIR by 2.5 points, one stitched finger separates “world class” from “double the national average.”

It creates underreporting incentives

When bonuses, contract prequalification, or plant rankings hang on TRIR, the cheapest way to improve the number is to not record the case: pressure the clinic toward first-aid-only treatment, reclassify restricted duty as "light tasks we'd have assigned anyway," or make reporting feel career-limiting. The number improves; the plant doesn't. The tell is an implausibly quiet log next to an empty near-miss system real floors generate signals, and a plant reporting neither injuries nor near misses isn't safe, it's silent. If you want a number people can't easily game, watch the ratio of near misses reported to recordables.

How should a plant actually use TRIR?

Report it, benchmark it against your own NAICS code, and trend it over multi-year windows, it's the language customers and insurers speak, and you need it for prequalification either way. But manage the plant on the leading side: near-miss volume and closure, JSA currency, inspection findings fixed on time. Those live or die on capture friction, which is why plants digitize them first, a near miss that takes a paper form and two signatures doesn't get reported; one that takes thirty seconds on the tablet already at the station does. That's the connected worker case in one sentence, and it's how Harmony approaches floor data generally: capture at the point of work, count what predicts, not just what already happened (see how it works).