Is the military's casualty (mortality) reporting system a reliable surrogate for International Classification of Diseases-coded death certificate information? To answer this question, the investigators compare official casualty data to the Air Force Mortality Registry for injury-related deaths occurring in 1991-1997. The investigators first derived International Classification of Diseases, Ninth Revision and Supplementary Classification of External Causes of Injury and Poisoning (E) codes for each death from casualty data and then compared the precision of those codes with the registry's E codes derived and medically coded from death certificates and autopsy reports. Sixty-five percent of registry E codes were "precise" vs. 35% from casualty data. The registry listed nonspecific E codes for 1% (12 deaths) vs. 6% (81 deaths) for casualty data. Unlike casualty data, the registry included expectant deaths that occurred within 120 days after medical retirement. The study concludes that casualty information compares poorly with that of the registry. Air Force Mortality Registry data should be used instead of casualty data for epidemiological research.