We have completed a case-control analysis of mesothelioma deaths among current and retired U.S. railroad employees. Cause-specific death certificates were obtained for 87% of 15,059 deaths reported by the railroad retirement board, and 20 mesotheliomas were identified according to death certificate diagnosis. A 10:1 matched analysis with railroad workers dying of nonmalignant, nonaccidental causes yielded a very strong association with prior railroad work in jobs with potential asbestos exposure (odds ratio = 7.2, 95% lower confidence limit = 3.3). Consideration of railroad occupations with regular asbestos exposures (e.g., skilled trades, steam locomotive repair) yielded an odds ratio of 21.4 (95% lower confidence limit = 8.7), but the occupations with potential intermittent exposure (e.g., engineers, firemen, carmen) yielded a nonsignificant odds ratio of 2.3 (95% lower confidence limit = 0.5). Applying mesothelioma mortality rates from this study to the population of U.S. railroad workers at risk yields an estimate of 416 cases of mesothelioma occurring among U.S. railroad workers between 1981 and 2000.