The ability of various data sources to identify fatal occupational injuries was examined. Death certificates, worker's compensation records, OSHA fatality reports, medical examiner records, and state department of health or labor reports from Wisconsin, Maryland, Michigan, Alaska, Colorado, California, New Jersey, Texas, Oklahoma, and Massachusetts were reviewed to identify all work related fatal injuries occurring in 1976, 1978, or 1980 through 1986. The available data sources for each state were linked and duplicate cases were eliminated to create a pool of fatal occupational injury cases. Capture rates, defined as the proportion of the pool identified by each separate data source, were computed. All ten states used death certificates to determine cases of occupational fatality. The capture rates ranged from 67 to 90%, average 81%. All states except Michigan and New Jersey used worker's compensation records to identify fatal occupational injuries. The capture rate ranged from 40 to 70%, mean 57%. All states except Michigan and Alaska used OSHA fatality reports to identify work related fatalities. The capture rates ranged from 21 to 42%, mean 32%. Maryland, New Jersey, and Oklahoma used medical examiner records to identify fatal work related injuries. The mean capture rate was 68%. Michigan, Texas, and Massachusetts used state department of health or labor records to record occupationally related fatalities. The mean capture rate was 27%. The authors conclude that using multiple sources of data will always result in obtaining a more complete source of work related fatal injuries than using one source alone. If only one source is used, death certificates will probably capture the largest proportion of work related fatal injuries.