Plain-language brief · Updated 2026-10-27

A low injury rate does not mean a low chemical release risk

A Spanish translation of this brief is available. English

What the numbers show

• Serious injuries happen behind clean records. We looked at 703 serious injuries at chemical plants where the plant's record for the year before was available. In 128 of them (18.21%), the plant had reported zero injuries the year before.
• The injury rate did not predict releases. Among plants with 100-249 employees, plants with a lower injury rate had a reportable chemical release the next year in 3.9% of plant-years. Plants with a higher rate: 3.4%. These are about the same.
• Release history did. A plant with a reportable release in the past five years had another the next year in 12.06% of plant-years. A plant without one: 1.47%. That is 8.2 times as often.
• The amount of chemicals held did too. Plants in the 5M+ lb group of regulated chemicals held had a release the next year in 8.12% of plant-years. Plants in the Under 100k lb group: 0.9%.

What this means for you

• Workers: a good injury record at a chemical plant does not show that the chemicals are well controlled. You can ask about past releases, the chemicals stored, and the emergency plan.
• Small employers: track chemical releases and near misses separately from injuries. Your injury count will not warn you about them.
• Neighbors and community groups: EPA release histories and chemical inventories are public. They say more about release risk than the injury rate.

Limits

These results are for chemical manufacturing plants in the United States, among 10,597 plants linked from public records. They are patterns in reported data. They do not show causes, and they are not a rating of any one plant. Serious injuries are counted only in states where federal OSHA covers private employers.

Where each number comes from

Each number is copied from a file in a public repository. The link opens that file.