Interpreting Employer Safety Grades

Why This Matters

OSHA enforcement data drives some of the highest-stakes decisions a worker makes: accepting an offer at a manufacturing plant, choosing between two warehouse employers, or deciding whether a small contractor is worth the risk. A raw DART rate of 4.2 sounds alarming in isolation but is actually below the industry mean for general warehousing. The same rate at an insurance carrier would be a four- alarm fire. Industry-normalized grading turns raw OSHA filings into a signal a non-specialist can actually use.

The challenge is that government data was designed for regulatory compliance and statistical reporting, not for the questions that most people are actually trying to answer. Understanding the gap between what the data measures and what you need to know is essential for drawing valid conclusions from PlainEmployers.

This guide walks through what each grade letter actually represents, how the underlying DART normalization works, which limitations to keep in mind (small-employer exemption, NAICS coarse-graining, reporting lag), and how to combine a safety grade with OSHA citation history when evaluating an offer at a real employer.

Key Concepts to Understand

The DART rate, defined: DART stands for Days Away, Restricted, or Transferred, the count of workplace cases that resulted in lost work time, restricted duty, or a job transfer, expressed per 100 full-time equivalent workers per year. OSHA publishes industry-average DART rates by NAICS sector each year, derived from establishment-level submissions to the Injury Tracking Application (ITA). When PlainEmployers shows an employer's DART, that number is the employer's own reported rate. When we assign a grade, we compare that rate against the OSHA- published industry mean.

How A–F grades map to industry-mean ratios: We use the ratio of the employer's DART to the industry's mean DART. An A means the employer rate is below 50% of the industry mean. B is 50–80%, C is 80–120% (i.e., within typical range), D is 120–180%, and F is above 180%. This banding is the same idea as a school grade curve: the question is not "is this number large" but "is this employer's rate well below, near, or far above what's normal for their kind of work."

Why some employers have no grade: OSHA's ITA reporting threshold is 250 employees in most industries and 20 employees in select high-hazard NAICS codes. Smaller employers below the threshold are not required to submit, so their DART rate is unknown to OSHA, and to us. Absence of a grade means the employer didn't have to report, not that the workplace is safe or unsafe. For unrated employers, OSHA enforcement records (citations, penalties) remain the available signal.

What the national DART spread actually looks like

Across 214,554 employers with a reported OSHA DART rate, the distribution below shows how injury rates are actually spread, before any industry normalization is applied.

OSHA DART rates across all reporting employers

Days-Away, Restricted, or Transferred cases per 100 full-time workers

employers

• 0.0–0.5
  87,317

  0.0–0.5

  87,317 employers

  100.0% of the leader · rank #1 · 87,317 employers · 41%

• 0.5–1.0
  13,384

  0.5–1.0

  13,384 employers

  15.3% of the leader · rank #2 · 13,384 employers · 6%

• 1.0–1.5
  14,512

  1.0–1.5

  14,512 employers

  16.6% of the leader · rank #3 · 14,512 employers · 7%

• 1.5–2.0
  13,526

  1.5–2.0

  13,526 employers

  15.5% of the leader · rank #4 · 13,526 employers · 6%

• 2.0–2.5
  11,875

  2.0–2.5

  11,875 employers

  13.6% of the leader · rank #5 · 11,875 employers · 6%

• 2.5–3.0
  10,276

  2.5–3.0

  10,276 employers

  11.8% of the leader · rank #6 · 10,276 employers · 5%

• 3.0–3.5
  8,731

  3.0–3.5

  8,731 employers

  10.0% of the leader · rank #7 · 8,731 employers · 4%

• 3.5–4.0
  7,634

  3.5–4.0

  7,634 employers

  8.7% of the leader · rank #8 · 7,634 employers · 4%

• 4.0–4.5
  6,525

  4.0–4.5

  6,525 employers

  7.5% of the leader · rank #9 · 6,525 employers · 3%

• 4.5–5.0
  5,532

  4.5–5.0

  5,532 employers

  6.3% of the leader · rank #10 · 5,532 employers · 3%

• 5.0–5.5
  4,603

  5.0–5.5

  4,603 employers

  5.3% of the leader · rank #11 · 4,603 employers · 2%

• 5.5–6.0
  3,877

  5.5–6.0

  3,877 employers

  4.4% of the leader · rank #12 · 3,877 employers · 2%

• 6.0–6.5
  3,271

  6.0–6.5

  3,271 employers

  3.7% of the leader · rank #13 · 3,271 employers · 2%

• 6.5–7.0
  2,700

  6.5–7.0

  2,700 employers

  3.1% of the leader · rank #14 · 2,700 employers · 1%

• 7.0–7.5
  2,396

  7.0–7.5

  2,396 employers

  2.7% of the leader · rank #15 · 2,396 employers · 1%

• 7.5–8.0
  2,096

  7.5–8.0

  2,096 employers

  2.4% of the leader · rank #16 · 2,096 employers · 1%

• 8.0–8.5
  1,764

  8.0–8.5

  1,764 employers

  2.0% of the leader · rank #17 · 1,764 employers · 1%

• 8.5–9.0
  1,522

  8.5–9.0

  1,522 employers

  1.7% of the leader · rank #18 · 1,522 employers · 1%

• 9.0–9.5
  1,279

  9.0–9.5

  1,279 employers

  1.5% of the leader · rank #19 · 1,279 employers · 1%

• 9.5–10.0
  1,189

  9.5–10.0

  1,189 employers

  1.4% of the leader · rank #20 · 1,189 employers · 1%

What this shows A raw DART rate here says nothing about safety on its own; the same rate can be an A grade in a high-risk sector and an F grade in a low-risk one, which is exactly why grades are computed against the industry mean, not this national curve.

Common Misconceptions

The most common misread of safety grades is comparing across NAICS sectors. A Grade A warehouse and a Grade A insurance carrier do not operate at the same absolute risk level, the carrier might have a DART of 0.5 against a 1.0 industry mean, while the warehouse runs at 3.0 against a 6.5 industry mean. Both are excellent for their kind of work, but the raw numbers are not interchangeable. Grades normalize across sectors; raw rates do not.

Another common mistake is assuming more recent data is always more relevant. Government data typically has a reporting lag of 12-24 months. The most recent available figures may describe conditions that have already changed, particularly in rapidly evolving sectors or regions. Always note the data vintage when making time-sensitive decisions.

A third misconception is that government data is always complete. In reality, reporting thresholds, voluntary participation rates, and processing delays mean that every dataset has gaps. PlainEmployers presents data as reported by source agencies, noting gaps where they are known. Absence of data does not mean absence of activity.

Practical Steps for Using the Data

Step 1, Start with the big picture. Before drilling into specific records, check the broad trends on PlainEmployers. What is the overall direction? Is the pattern you are investigating part of a larger trend or an isolated anomaly?

Step 2, Compare appropriately. When evaluating any specific data point, compare it against similar entities rather than the national average. Geographic, industry, and size differences create natural variation that makes broad comparisons potentially misleading.

Step 3, Check the source documentation. Every data point on PlainEmployers traces back to a government source. When the stakes are high, career decisions, policy analysis, research publications, verify critical figures against the primary source. We provide source attribution on our data pages and about page.

Step 4, Apply judgment that data cannot provide. Data is a starting point, not a final answer. The best decisions combine quantitative data with qualitative context, local knowledge, expert consultation, and direct observation. Use PlainEmployers data to narrow your focus and inform your questions, not to replace professional judgment or lived experience.

Frequently Asked Questions

What data does PlainEmployers use?

PlainEmployers uses data from DOL H-1B filings, OSHA safety records, and WARN Act notices. All data comes from public sources and is processed through our pipeline for searchability and analysis.

How often is the data updated?

We update our database as new data becomes available from source agencies. Frequency depends on the source release schedule, which varies from monthly to annually depending on the dataset.

How should I interpret the data?

Always compare within appropriate reference groups. Aggregate statistics describe populations, not individual cases. See our full guide library for detailed interpretation frameworks.

Is PlainEmployers free to use?

Yes. PlainEmployers is completely free, requires no account, and is supported by non-intrusive advertising. We believe public data should be freely accessible to everyone.