Background. Fatal occupational injury rates have been declining since the 1970s in most industrialized countries. Studies in some areas suggest that not all workers have enjoyed equal improvements in safety, however. We conducted a study to characterize trends in fatal occupational injury rates in the United States and identify factors associated with the magnitude of improvement in risk. Methods. All deaths from injury at work in the US during 1980-1996 were identified from the National Traumatic Occupational Fatality data base maintained by the National Institute for Occupational Safety and Health; the size of the populations at risk was estimated from the Census of Population for 1980 and 1990. Mortality rates were computed for three categories of injury deaths: unintentional injuries, homicides, and all injuries combined. The annual rate of change was estimated by using Poisson regression to model the death rate as a function of time. Results. The overall rate of fatal occupational injuries declined by 3% per year from a baseline of 5.8 per 100,000 in 1980. The rate of homicide declined at a slower rate, <1% per year. The rate of improvement was significantly faster for men (3%/year) than for women (1%/year), although the absolute risk remained much higher for men. The rate of death decreased rapidly among workers younger than 20 (7%/year), while workers over age 50 experienced a slower decline (2%/year). Workers of European ancestry enjoyed more rapid improvement in risk (about 4%/year) than African-:Americans (3%/year) or workers of other ethnic groups (< 1 %/year). Injury rates for most occupations declined at a rate near the average of 3%/year, but for some, including farmers, fishers, machine operators, and machine repairers, there was essentially no change. Several important industries, including transportation, mining and chemicals, experienced declines of 4-6%/year, but a number of others, including taxi services, textiles, and metal manufacturing, saw no change or had increasing fatality rates. The rate of homicide also increased in a number of occupations and industries. Conclusions. Overall, US workers enjoyed steadily improved safety during the 1980s and 1990s. Possible explanations include successful safety programs, technological changes, and elimination or export of hazardous operations through economic restructuring. The change was not uniform, however. Fatality rates increased in some occupations and industries, and there was lime change in the rate of homicide. Further research on the determinants of these trends is needed.