TRIR counts every OSHA recordable case. DART counts only the cases that involved days away from work, restricted work or a job transfer — columns H and I of the 300A. Same formula, same 200,000-hour base, smaller numerator. Nationally in 2024, private industry ran a TRIR of 2.3 and a DART of 1.4.
The difference, in columns
Both rates come off the same summary you post every February, and both use the same arithmetic: cases × 200,000 ÷ hours worked. What separates them is which boxes you add up.
| 300A column | What it counts | TRIR | DART |
|---|---|---|---|
| G | Total number of deaths | Yes | No |
| H | Cases with days away from work | Yes | Yes |
| I | Cases with job transfer or restriction | Yes | Yes |
| J | Other recordable cases | Yes | No |
| K, L | Total days away, and days of transfer or restriction | No | No |
Two consequences fall straight out of that table. A fatality does not raise your DART rate — it sits in column G. And days never enter either rate: a case that cost 180 days off counts exactly the same as one that cost two. Days are what the severity rate is for.
Run both from the same case counts in the DART rate calculator and the TRIR calculator; the severity and lost time rate calculator handles the day-based rates.
The judgment call that moves a case
Whether a case lands in column I or column J is not a clerical detail — it is the decision that sets your DART rate. 29 CFR 1904.7(b)(4) defines it, and the definition is narrower than most people assume.
Restricted work occurs when, as the result of a work-related injury or illness, you keep the employee from performing one or more of the routine functions of his or her job, or from working the full workday that he or she would otherwise have been scheduled to work.29 CFR 1904.7(b)(4)(i)
Six rules decide the rest, and each one is a place real cases go wrong:
- Routine functions are the activities the employee performs at least once per week — 1904.7(b)(4)(ii). A restriction that only stops something the employee does twice a year is not a restricted work case.
- The day of the injury does not count — 1904.7(b)(4)(iii). Restriction or transfer imposed only for that day is not recorded as restricted work at all.
- A partial shift is a day of restriction — 1904.7(b)(4)(v). Sending someone home two hours early the day after the injury puts the case in column I.
- Producing less is not a restriction — 1904.7(b)(4)(vi). If the employee performs all routine functions and works the full shift, the case is not restricted work even if output drops.
- A vague restriction defaults to recordable — 1904.7(b)(4)(vii). Ask whether the employee can do all routine functions and work the full shift. If you cannot get an answer, record it as restricted work.
- A written restriction counts even if the employee ignores it — 1904.7(b)(4)(viii). If a health care professional recommends a restriction meeting the definition and the employee does the full job anyway, it is still a restricted work case — and you should be enforcing the restriction.
Light duty does not lower your DART
This is worth saying plainly because entire return-to-work programs are sold on the opposite claim. Bringing an injured worker back on modified duty moves the case from column H (days away) to column I (restriction or transfer). Both columns sit inside DART. The rate does not move by a single decimal.
| Measure | Effect of moving a case to light duty |
|---|---|
| TRIR | No change — the case was already recordable |
| DART rate | No change — H and I are both counted |
| Lost time case rate (column H only) | Falls |
| Severity rate / days lost (columns K and L) | Falls, often sharply |
| Workers' compensation indemnity cost | Falls |
Modified duty is still worth doing: people recover better at work than at home, the day counts fall and the claim costs fall with them. Just do not promise a customer that it will bring your DART down, because it will not, and the 300A they ask to see will show it.
Why OSHA targets on DART, not TRIR
DART is a crude severity filter. It drops the cases that ended at the clinic with a prescription and no time away — real cases, but weaker evidence of a dangerous workplace. That is why OSHA's programmed inspection plan for general industry is built on it.
The most recent published Site-Specific Targeting directive, CPL 02-01-064 of February 7, 2023 (now marked as archived on OSHA's site), generates inspection lists in four ways from the Form 300A data establishments submit electronically:
- High-rate establishments — with separate DART thresholds for manufacturing (NAICS 31-33) and for non-manufacturing, because average rates differ so much between them.
- Upward-trending establishments — those at or above twice the private-sector national average in the base year whose rates kept climbing.
- Low-rate establishments — a random sample, specifically to verify data accuracy and quality. Reporting a suspiciously clean number is itself a selection criterion.
- Non-responders — a random sample of establishments that failed to submit their 300A data at all.
Construction is excluded from the SST plan — it is covered by other programs. And note the third bullet: under-recording is not a way out of the list. It is a way onto it.
Reading the gap between your two rates
The national picture gives you a reference for the shape, not a target. In 2024, private industry in the United States looked like this:
| Rate | 2024 | 2023 | 2019 |
|---|---|---|---|
| Total recordable cases (TRIR) | 2.3 | 2.4 | 2.8 |
| DART | 1.4 | 1.5 | 1.5 |
| Days away from work only | 0.8 | 0.9 | 0.9 |
So about six recordable cases in ten are DART cases, and a little over half of those involve actual days away. Against that, two patterns are worth investigating in your own numbers:
- DART close to TRIR. Either your recordables are genuinely severe, or column J is being under-recorded — the case that ended in stitches or a prescription with no time off is the one most often missed.
- DART far below TRIR. Either your injuries really are minor, or restricted work is not being captured: partial shifts, informal "take it easy for a few days", and unwritten accommodations that never reach the recordkeeper.
Either way, the diagnosis is in the 300 Log, not in the rate. Read the case descriptions.
Bottom line
- TRIR = columns G + H + I + J. DART = columns H + I. Same formula, same 200,000 base.
- Fatalities are in TRIR and not in DART; days are in neither.
- Restricted work has a precise definition in 1904.7(b)(4), and the day of the injury never counts.
- Light duty moves a case from H to I. It lowers days lost and claim costs; it does not lower DART.
- OSHA's Site-Specific Targeting plan selects on DART — high, rising, and implausibly low.