DART rate (days away, restricted or transferred): the measure United States employers calculate from the OSHA 300 Log

DART is usually introduced as a severity filter on top of the recordable count, and that description is not wrong so much as incomplete. The more useful way to hold it is this: DART is the one common safety metric that is deliberately indifferent to whether an employer sends an injured worker home or puts them on modified duty. Lost-time rates move sharply when a company opens a light duty program. DART does not. That indifference is the whole design, and it means the interesting signal is not the DART number itself but the split inside it.

What is the DART rate?

DART stands for days away, restricted, or transferred. It counts the recordable cases whose consequence was serious enough to interrupt the work: the employee could not come in, or came in but could not do the job as normal.

BLS defines the DART population as “cases with days away from work, job transfer or restriction, or both (DART) (Column H + Column I)” on the Form 300A summary. Those two columns are mutually exclusive by regulation. A case with any days away goes in column H; column I takes cases involving “restricted work or job transfer but does not involve death or days away from work” (29 CFR 1904.7(b)(4)). No case is ever counted twice.

One point catches people out. DART counts cases, not days. An employee off for four months and an employee restricted for two days each contribute exactly 1 to the numerator. The day counts are recorded in separate columns on the 300 Log and play no part in the DART calculation at all.

What is the DART formula?

DART rate = (Cases with days away, restricted work or job transfer x 200,000) / Total hours worked by all employees

The 200,000 base and the hours-worked rules are the same ones the Total Recordable Incident Rate uses, and that page sets out where the base comes from and what counts as an hour actually worked. BLS works its own DART example with a company logging 3 qualifying cases against 400,000 hours: “(3 x 200,000) / 400,000 = 1.5” (BLS). Because the base and the denominator are identical to TRIR’s, DART can never exceed TRIR at the same establishment.

Which cases count toward DART?

A case qualifies through days away, through restricted work, or through transfer to another job. It has to be a recordable case first, so the question of what makes a case recordable, and the separate question of what has to be reported to OSHA rather than recorded, both sit upstream of this calculation.

Days away is the simpler branch, though its counting rules surprise people. You begin counting “on the day after the injury occurred or the illness began”, and you count calendar days rather than scheduled days: weekends, holidays and vacation days are included “if the employee would not have been able to work on those days” (29 CFR 1904.7(b)(3)). There is a cap, and employers may “cap the total days away at 180 calendar days”.

Restricted work is the branch where judgment lives. It occurs when the employer keeps the employee “from performing one or more of the routine functions of his or her job, or from working the full workday”, or a physician or other licensed health care professional recommends the same. Routine functions are defined narrowly as “those work activities the employee regularly performs at least once per week”. Three exclusions matter in practice: a restriction imposed only for the day of the injury is not recordable as restricted work; working a partial shift counts as a day of restriction, except on the day of injury; and producing less output while still performing every routine function is not restricted work at all.

The regulation also resolves ambiguity in one direction. Where a restriction is vague, the employer should ask the clinician whether the employee can do all routine functions and work a full shift, and if that clarification cannot be obtained, “record the injury or illness as a case involving restricted work” (29 CFR 1904.7(b)(4)(vii)). The default is to count it. An organization that chases clinicians for clarification will therefore report a lower DART than an identical organization that does not, from identical injuries.

A worked example you can check

Take the same establishment used on our TRIR page: one manufacturing site, 140 employees, calendar year 2025, 268,400 hours actually worked. Of its 9 recordable cases for the year, 2 sat in column H as cases with days away from work and 3 in column I as cases with job transfer or restriction.

DART cases = column H + column I = 2 + 3 = 5

DART rate = (5 x 200,000) / 268,400 DART rate = 1,000,000 / 268,400 DART rate = 3.73

The TRIR for the same site is (9 x 200,000) / 268,400 = 6.71. So 5 of 9 recordable cases, about 56%, were serious enough to interrupt work.

That ratio is the number worth writing down. For private industry in 2024, BLS published a total recordable case rate of 2.3 per 100 full-time equivalent workers, made up of 0.8 for cases with days away from work and 0.5 for cases with days of job transfer or restriction (BLS Table 1). Those rates are rounded to one decimal place, but they imply a DART component of roughly 1.3 against a total of 2.3, or about 57%. The example site’s severity mix is therefore close to the national private industry mix even though its overall rate is nearly three times as high. Two quite different stories, and only one of them is visible in either rate on its own.

Those national figures are survey estimates rather than a census, and the survey’s scope is narrower than American industry as a whole. Our TRIR page sets out what the published rates can and cannot be benchmarked against.

How does DART differ from TRIR?

TRIR counts every recordable case. DART counts the subset that involved days away, restriction or transfer. The excluded group is the “other recordable cases” in column J: cases that met the recording criteria through medical treatment beyond first aid, loss of consciousness, or a significant diagnosis, but where the employee kept working normally.

There is one exclusion nobody expects. Deaths sit in column G, and the DART population is column H plus column I. A fatality raises an employer’s TRIR and leaves its DART rate untouched. DART is a measure of work interruption, not of harm, and at the extreme of harm there is no work to interrupt.

Is TCIR the same as TRIR?

Effectively, yes. TCIR stands for Total Case Incident Rate and describes the identical calculation: all recordable cases, multiplied by 200,000, divided by hours worked. Some contractor prequalification portals prefer TCIR, others TRIR. None of the acronyms is official, for the reasons set out on the TRIR page. If you are asked for a TCIR, supply your TRIR and say so.

How can a company have a low TRIR and a poor DART?

Because the two rates answer different questions, and an organization can do well on frequency while doing badly on severity.

Picture two firms, each working 400,000 hours in a year. Firm A records 12 recordable cases, of which 2 involved days away or restriction. Firm B records 5, of which 4 did.

• Firm A: TRIR = (12 x 200,000) / 400,000 = 6.00. DART = (2 x 200,000) / 400,000 = 1.00.
• Firm B: TRIR = (5 x 200,000) / 400,000 = 2.50. DART = (4 x 200,000) / 400,000 = 2.00.

Firm B’s TRIR is less than half of Firm A’s. Its DART rate is double. On a prequalification form that asks only for TRIR, Firm B looks like the safer contractor, and on the measure that tracks whether people were actually stopped from working it is twice as bad.

This is not a contrived case. It appears in the national data. In 2024, construction (NAICS 23) recorded a lower total recordable case rate than manufacturing (NAICS 31-33), at 2.2 against 2.7, but a higher rate of cases with days away from work, at 0.9 against 0.8 (BLS Table 1). Days away cases were about 41% of construction’s recordables and about 30% of manufacturing’s. Construction records fewer cases per hour worked and a greater share of them are the ones that put someone out of work. A buyer comparing sectors on TRIR alone would draw precisely the wrong conclusion about where the serious harm is.

Does a light duty program lower DART?

No, and this is the most persistent misunderstanding about the metric.

When an employer runs a structured modified-duty program, a case that would have produced days away instead produces restricted work. Under 29 CFR 1904.7(b)(4) that case moves from column H to column I. Both columns are inside DART, so the DART rate does not move. The TRIR does not move either, because the case was recordable in both scenarios. What moves is the days away from work rate and any lost-time measure, sometimes dramatically.

That is the design working as intended. DART was constructed to be robust to a decision that is clinically and commercially legitimate but that renders lost-time rates incomparable between employers. It also means the split within DART, the ratio of column H to column I, is a reasonable proxy for how developed an employer’s return-to-work capability is, whereas the DART total is not. That same decision has a price attached elsewhere: getting an injured employee back to productive work sooner tends to hold down the eventual valuation of the claim, which is what reaches an employer’s experience modification rate two years later.

What DART does not tell you

DART is silent on duration, so a case that cost 170 days and a case that cost 2 days are the same case to it. It is silent on fatalities. It is silent on cause. And like every rate built from the 300 Log, its numerator depends on injuries being reported and classified consistently, which is a property of the organization rather than of the work, and which the TRIR page treats at length.

Used together the pair is genuinely informative: TRIR for how often, DART for how often it mattered, and the ratio between them for whether your severity mix is drifting. Used singly, either can be made to flatter.

Related terms

• TRIR, the Total Recordable Incident Rate: the total case measure DART is derived from, and where the 200,000 base, the hours rules and the benchmarking cautions are set out.
• Recordable versus reportable: which events go on the log and which have to be phoned in to OSHA.
• Experience modification rate: what the same injuries cost once they reach a workers compensation premium.
• OSHA recordkeeping: the 300 Log, the 300A annual summary and the 301 incident report, and where columns G, H, I and J come from.

Frequently asked questions

Sources

• 29 CFR Part 1904, Recording and Reporting Occupational Injuries and Illnesses
• 29 CFR 1904.7, General recording criteria, including (b)(3) days away and (b)(4) restricted work and job transfer
• 29 CFR 1904.29, Forms
• Bureau of Labor Statistics, How to compute a firm’s incidence rate
• Bureau of Labor Statistics, Handbook of Methods, SOII concepts
• Bureau of Labor Statistics, Employer-Reported Workplace Injuries and Illnesses, 2024, released 22 January 2026
• Bureau of Labor Statistics, Table 1, incidence rates by industry and case type, 2024

Last reviewed: 16 September 2026

About Logincident. Logincident is a data and software company whose configurable platform captures structured evidence at the point an event happens and presents it in dashboards and reports. We are not a law firm or a claims handler, and nothing on this page is legal advice.