The accuracy of the injury at work item on death certificates for surveillance of occupational injury deaths was investigated. Death certificates, OSHA fatality/catastrophe reports, workers' compensation reports, and medical examiner reports were used to identify 329 Oklahoma workers who were fatally injured between 1985 and 1986. An overall annual average fatal injury rate of 11.3 deaths per 100,000 workers per year (d/w/yr) (2.3d/w/yr for females and 18.7d/w/yr for males) was found. Certificates identified 93% of the female deaths and 72% of the male deaths. Of the male deaths, 82% were identified by medical examiner reports, 57% through workers' compensation reports, and 26% through OSHA reports. Death certificates missed 57% of the fatalities occurring in helpers and laborers, transportation and material moving workers, farmers, precision production workers, crafters, and repairers. The authors confirm that no single data source contains all the death information that is needed to describe fatal strategies; however, they suggest that greater representativeness may be achieved by combining all data sources.