The two formulas, written out
TRIR = (recordable cases × 200,000) ÷ hours worked
DART = (cases with days away, restricted duty or transfer × 200,000) ÷ hours worked
A worked example. A 40-person outfit whose crews put in 84,000 hours, with three recordable cases and one of them costing days away, has a TRIR of 7.14 and a DART of 2.38.
Where the 200,000 comes from
It is not a magic number and it is not a threshold. It is 100 full-time workers, 40 hours a week, 50 weeks a year. Multiplying by it converts your rate into "cases per 100 full-time workers per year," which is the only reason a 12-person company and a 900-person company can be compared at all.
That also explains why a small company's TRIR swings so hard. At 84,000 hours, one recordable case moves your TRIR by 2.38. At 840,000 hours it moves it by 0.24. On a small crew the number is mostly telling you how many people got hurt, not how your program is doing, and anybody reading it seriously knows that.
The rule does not set this number
Worth being clear about, because a lot of pages blur it. The OSHA recordkeeping rule, 29 CFR Part 1904, decides which cases go on your log and in which column. It does not set a TRIR, it does not publish an acceptable one, and there is no federal penalty for a high rate on its own. The 200,000-hour arithmetic is the convention used on OSHA's own 300A worksheet and in the Bureau of Labor Statistics injury statistics, and everybody who asks you for a rate is using it.
So the number matters for a different reason: a hiring client's contract, or a prequalification network's grade, or an insurance conversation. Those set the bar, not the regulation.
The two things that make the number wrong
1. The case count does not match the 300 log
This is the common one, and it is the one a coordinator catches. Somebody fills in the prequalification form from memory in March, and the number does not agree with the 300A that was posted in February. Now two documents from the same company disagree about the same year, and the conversation stops being about safety.
Fix it at the source: get the recordability call right when the incident happens, not eleven months later. That is what the free recordability checker is for.
2. The hours are estimated
Hours worked is the denominator, so an estimate there moves the answer more than anything else on the page. Payroll knows the real figure. Use it.
What counts as a "good" TRIR
There is no legal answer, and anybody who gives you a single number without asking what you do is guessing. The comparison that means something is your own industry's published rate for the same year, which the Bureau of Labor Statistics puts out by industry code, and the cap written into the contract or the prequalification network in front of you. Read that cap before you read anybody's benchmark.
What this page does not do
It does not decide whether a case is recordable. That is a separate decision under Part 1904 with its own tests, and getting it wrong changes the numerator here. The checker walks that decision question by question and cites the paragraph behind every answer.