In 2024, U.S. private-industry employers recorded about 2.5 million nonfatal workplace injuries and illnesses — the lowest total in more than two decades — and yet that number is the real story of workplace harm, not the 5,070 deaths that make headlines. The BLS Survey of Occupational Injuries and Illnesses (SOII), released January 22, 2026, is the companion to the fatality census: where the CFOI counts who died, the SOII counts who got hurt. Put the two together and you get the single most useful mental model in safety — the injury pyramid — plus a cautionary lesson about reading a "record low" too quickly.
At a glance (BLS SOII, 2024): ~2.5 million recordable injuries & illnesses · down 3.1% from 2023 · lowest since 2003 · rate 2.3 per 100 FTE (from 2.4), also a series low · illnesses 148,000 (down 26%) · ~493 recorded injuries per workplace death. Source: BLS Survey of Occupational Injuries and Illnesses.
Key findings — U.S. nonfatal work injuries, 2024
- The base of the pyramid is 2.5 million. For every one of the 5,070 workers killed in 2024, about 493 more were hurt badly enough to be recorded — and countless minor injuries were never logged at all.
- Both the count and the rate hit series lows. 2.5 million cases is the fewest since the series began in 2003; the 2.3 rate per 100 FTE is the lowest too.
- Part of the 2024 drop is a mirage. A 26% fall in illnesses — led by a 46% collapse in respiratory cases — is the last of pandemic-era COVID reporting washing out, not a safety gain.
- The injury map differs from the death map. Construction leads deaths, but health care and social assistance reports the most injuries of any sector, and transportation/warehousing carries some of the highest rates.
- This data is your 300 logs, aggregated. The SOII is built from employer OSHA 300/301 records under 29 CFR Part 1904 — the national picture is only as accurate as each worksite's recordkeeping.
The injury pyramid: 493 to 1
Safety professionals have talked about the "accident pyramid" for a century — the idea that every fatality sits atop a much larger base of serious injuries, minor injuries, and near misses. The 2024 BLS data lets you put real national numbers on the top two layers.
The 2024 workplace-harm pyramid
Two BLS programs, one structure · not drawn to linear scale (a true-scale tip would be invisible)
≈ 493 recorded injuries for every death
Sources: BLS CFOI 2024 (5,070 deaths) and BLS SOII 2024 (~2.5M nonfatal recordable cases). 2,500,000 ÷ 5,070 = 493. Unrecorded minor injuries and near misses widen the true base further.
That 493-to-1 ratio is why a safety program built only around fatality prevention is incomplete. The 5,070 deaths are catastrophic and non-negotiable — but they are statistically rare events sitting on an enormous base of sprains, lacerations, fractures, and burns that drain payroll, morale, and productivity every single day. A program that drives down the base — the near misses and first-aid cases — is the one that eventually starves the tip. That is the entire logic behind near-miss reporting and job hazard analysis.
A genuine record low — read on the rate, not the count
The 2.5 million headline is the lowest since the modern series began in 2003. But as with fatalities, the count alone can mislead, because it doesn't adjust for how many people are working. The honest measure is the rate — recordable cases per 100 full-time-equivalent workers — and it, too, hit a series low of 2.3 in 2024, down from 2.4.
Private-industry recordable injuries & illnesses, 2020–2024
Total cases (millions) · bars scaled to the 2022 peak
Source: BLS SOII (private industry). 2020 = 2,654.7K · 2021 = 2,607.9K · 2022 = 2,804.2K · 2023 ≈ 2.58M (down 8.4% from 2022) · 2024 ≈ 2.50M (down 3.1%). 2024 is the lowest since 2003; rate 2.3 per 100 FTE.
The trend is real and it is good: across two decades, American work has become measurably less injurious per hour worked. But 2024 has an asterisk.
The illness mirage: why 2024's drop is partly a statistical artifact
Dig into what fell in 2024 and a large share of the improvement is not fewer injuries — it is fewer illnesses, and specifically fewer respiratory illnesses.
What actually drove the 2024 decline
The headline fall was disproportionately an illness story, not an injury story.
Source: BLS SOII 2024. Respiratory-illness reporting spiked in 2020–2022, when large numbers of work-related COVID-19 cases met the recording criteria (a work-related illness is recordable under 29 CFR 1904.7(a)); the fall in those cases pulls the 2024 totals down for reasons unrelated to physical-injury safety.
This is the analytical trap. If your board sees "injuries down 3.1%" and concludes the safety program is working, they may be reading a demographic and epidemiological artifact — the fall in recorded COVID-19 cases — as an operational win. Strip out the illness swing and the underlying injury picture is flatter than the headline suggests. A disciplined EHS leader reports the injury rate and flags what moved it, rather than banking a mirage.
Two maps: where workers die vs. where workers get hurt
The most useful thing the SOII does is show that the injury map is not the death map. The hazards that kill and the hazards that injure concentrate in different industries.
Deadliest vs. most-injured — two different rankings
Fatalities (CFOI) and nonfatal injuries (SOII) do not rank industries the same way.
| Lens | What tops the list | Why it matters |
|---|---|---|
| Deaths (CFOI) | Construction is the deadliest industry; transportation is the deadliest event. | High-energy, low-frequency events — falls, struck-by, crashes. |
| Most injuries (SOII) | Health care & social assistance reports the most cases of any sector (about 562,500 in 2023). | Overexertion, patient handling, slips — frequent, rarely fatal. |
| Highest rates (SOII) | Transportation & warehousing runs hot — couriers & messengers hit 9.4 per 100 in 2022. | Ergonomics and pace, not catastrophic hazards. |
Sources: BLS CFOI 2024 (fatalities) and BLS SOII (nonfatal). Sector figures are dated where noted; 2024 industry detail is published in the SOII industry tables. Health-care 562,500 is the 2023 sector total, historically the largest of any sector.
The implication for a multi-site or multi-trade employer is direct: your fatality risk and your injury risk may live in different places. A general contractor's deaths cluster in leading-edge and struck-by work; its injuries cluster in strains, hand lacerations, and slips. Manage both — one with engineered fall/struck-by controls, the other with ergonomics, housekeeping, and PPE — and log them correctly, because the log is where the data starts. A strain treated only with hot or cold therapy is first aid and stays off the log (29 CFR 1904.7(b)(5)(ii)(E)).
The recordkeeping connection: this data is your 300 log
What most readers miss: the SOII is not an independent measurement. It is built directly from the OSHA 300 logs and 301 incident reports that covered employers must keep under 29 CFR Part 1904. BLS samples those employer records and weights them into the national estimate. Which means:
- Whether a case is a recordable injury at all is decided at each worksite by the general recording criteria in 1904.7 — death, days away, restricted work, transfer, medical treatment beyond first aid, loss of consciousness, or a significant injury or illness diagnosed by a physician or other licensed health care professional (29 CFR 1904.7(a), (b)(7)).
- The DART rate that insurers and prime contractors judge you on comes straight off that same log.
- Miscount there — over- or under-record — and you distort both your own metrics and, in aggregate, the national picture.
Accurate recordkeeping is therefore not clerical busywork; it is the foundation of every injury statistic, every experience-modification rate, and every prequalification you will ever submit. Get the recordability decision right, keep the 300 log clean, and file the 301 for each case.
Build the base of the pyramid into your program
The 2.5-million base is where a program is actually won or lost — and it starts with clean, defensible records. HazComFast keeps your OSHA logs, incident reports, and training files audit-ready, and a near-miss logged without a signal sends itself the moment you're back online. Decide recordability, cost the exposure, then close the gaps:
The recordkeeping rules behind every number in this survey
| What decides the count | Rule |
|---|---|
| Work-related: an event or exposure at work caused or contributed to the condition, or significantly aggravated a pre-existing one | 29 CFR 1904.5(a) |
| Recordable: death, days away, restricted work or transfer, medical treatment beyond first aid, loss of consciousness, or a significant diagnosed injury or illness | 29 CFR 1904.7(a) |
| First aid, not recordable: the closed list of treatments | 29 CFR 1904.7(b)(5)(ii) |
| Days away counted in calendar days, with an optional cap at 180 | 29 CFR 1904.7(b)(3)(vii) |
| Restricted work: kept from routine functions or a full workday | 29 CFR 1904.7(b)(4)(i)(A) |
| Each case entered within 7 calendar days | 29 CFR 1904.29(b)(3) |
| Privacy concern cases entered without the employee's name | 29 CFR 1904.29(b)(6) |
| Annual summary posted from February 1 to April 30 | 29 CFR 1904.32(b)(6) |
| Records kept for five years after the year they cover | 29 CFR 1904.33(a) |
| Companies with 10 or fewer employees partially exempt, unless OSHA or BLS asks in writing | 29 CFR 1904.1(a)(1) |
| An employer selected by BLS completes and returns the survey form | 29 CFR 1904.42(a) |
| Even an exempt employer, and one in a State-Plan state, answers when BLS asks | 29 CFR 1904.42(b)(3)-(4) |
| Electronic submission of 300A, 300, and 301 data for designated establishments | 29 CFR 1904.41(a)(1)-(2) |
| Deaths reported within 8 hours, hospitalizations within 24 | 29 CFR 1904.39(a)(1)-(2) |
| No discharge or discrimination for reporting an injury | 29 CFR 1904.35(b)(1)(iv) |
Methodology and limits
- Source and vintage. All nonfatal figures are from the BLS Survey of Occupational Injuries and Illnesses, 2024 reference year, released January 22, 2026 (USDL-26-0101) — the most recent available. Unlike the CFOI census, the SOII is an employer survey (a large probability sample of establishments), so figures are estimates with sampling error, not a complete count.
- Private industry. The headline 2.5 million and the 2.3 rate are private industry; state and local government are surveyed separately and add to the national total.
- The pyramid ratio is cross-program. The ~493:1 figure divides the SOII nonfatal count (2.5M) by the CFOI fatal count (5,070). They are different programs with different methods; the ratio is an illustrative order-of-magnitude, not a precision statistic.
- Industry specifics are dated. Where a sector figure predates 2024 (health care 562,500 = 2023; couriers 9.4 = 2022), it is labeled; the full 2024 industry detail lives in the BLS SOII industry tables.
- Percentages are arithmetic on the published counts.
Manage the base of the pyramid: record it, then reduce it
The fatality census tells you work can kill; the injury survey tells you how work actually wears people down — 2.5 million times in 2024. The record-low rate is real and worth celebrating, but only if you read it honestly: net of the COVID-illness unwind, and with the understanding that the 2.5-million base, not the 5,070-death tip, is where most safety programs are decided. Manage the base — record it accurately, analyze it, and drive it down — and the tip takes care of itself.
Related data studies
- The fatality side: What 5,070 Workplace Deaths in 2024 Reveal (CFOI) · Why Falls Stay OSHA's #1 Construction Hazard
- Recordkeeping: OSHA 300 Log Recordkeeping Guide · When Is an Injury Recordable? · Forms 300/300A/301 · ITA Electronic Submission
- Cost & enforcement: The True Price of a Construction Fall · OSHA Penalties by State
Sources & verification (verified 2026-07-13): All nonfatal figures from the BLS Survey of Occupational Injuries and Illnesses, 2024 reference year, released January 22, 2026 (USDL-26-0101): about 2.5 million recordable cases (down 3.1% from 2023), lowest since 2003; total recordable rate 2.3 per 100 FTE (from 2.4), also a series low; total illnesses 148,000 (down 26.0%), respiratory illnesses 54,000 (down 46.1%, lowest since 2019). Count series (private industry): 2020 = 2,654.7K, 2021 = 2,607.9K, 2022 = 2,804.2K, 2023 ≈ 2.58M (BLS: 2.6M, down 8.4% from 2022), 2024 ≈ 2.50M. Fatality figure (5,070) from the BLS Census of Fatal Occupational Injuries, 2024. Pyramid ratio 2,500,000 ÷ 5,070 = 493. Industry figures dated where noted (health care & social assistance 562,500 = 2023; couriers & messengers 9.4 per 100 = 2022). SOII is an employer survey (estimates with sampling error); the SOII is compiled from OSHA 300/301 records under 29 CFR Part 1904. Not legal advice.
Frequently Asked Questions
How many workplace injuries were there in the US in 2024?
Private-industry employers reported about 2.5 million nonfatal workplace injuries and illnesses in 2024, according to the BLS Survey of Occupational Injuries and Illnesses released January 22, 2026. That is down 3.1% from 2023 and the lowest total in the series going back to 2003. The total recordable case rate fell to 2.3 cases per 100 full-time-equivalent workers, from 2.4 in 2023 — also a series low. Each of those cases met OSHA's general recording criteria (29 CFR 1904.7(a)).
What is the difference between the CFOI and the SOII?
They are two different BLS programs. The Census of Fatal Occupational Injuries (CFOI) is a complete count of workers killed on the job — 5,070 in 2024. The Survey of Occupational Injuries and Illnesses (SOII) is an employer survey of NONFATAL recordable injuries and illnesses — about 2.5 million in 2024. Together they describe the whole pyramid: a small tip of deaths sitting on a broad base of injuries. An employer selected for the SOII must complete and return the BLS survey form (29 CFR 1904.42(a)).
How many injuries occur for every workplace death?
Roughly 493. Dividing the 2.5 million nonfatal recordable cases in the 2024 SOII by the 5,070 deaths in the 2024 CFOI gives about 493 recorded nonfatal injuries and illnesses for every worker killed. Because many minor injuries are never recorded at all, the true ratio of all harm to each death is larger still. This is the statistical backbone of the safety pyramid: fatalities are the visible tip of a very broad base. A case is recorded only when it meets the general criteria, so injuries treated with first aid alone never reach the count (29 CFR 1904.7(a), (b)(5)(ii)).
Did workplace injuries really get safer in 2024, or is the drop misleading?
Both. The total recordable rate of 2.3 per 100 FTE is a genuine series low, so the long-run trend is real. But a large part of the 2024 drop came from a 26% fall in illness cases — driven by a 46% collapse in respiratory illness cases as the last of the pandemic-era COVID reporting washed out. That part of the decline is a return to baseline, not a safety improvement. Read the injury rate, and read it net of the illness swing. The rule did not change: a work-related illness is recordable under the same criteria as an injury (29 CFR 1904.7(a)), so the swing reflects fewer recorded cases.
Where does the SOII injury data come from?
From employers' own OSHA recordkeeping. The SOII is built from the OSHA 300 logs and 301 incident reports that covered employers are required to keep under 29 CFR Part 1904, sampled and weighted by BLS. In other words, the national injury statistics are only as good as the individual 300 logs behind them — which is why accurate, defensible recordkeeping matters at every single worksite.
OSHA figures and citations here come from our regulatory source-of-truth modules, last checked against the eCFR, OSHA.gov, and the Federal Register on October 5, 2026. Last reviewed October 6, 2026.
About This Article
Published by: HazComFast
Published: July 13, 2026
Last Updated: October 6, 2026
This content is for informational purposes only and does not constitute legal advice.