A mortality study of 76,160 men who served on US nuclear submarines is reported. Indirect standardization was used to compare mortality rates to those of the US male population. Multiplicative models were developed to explore patterns of mortality within the cohort. Mortality rates for leukemia, acute myocardial infarction, and for motor vehicle accidents were equivalent to those of US males; rates for other causes were lower, generally consistent with the "healthy worker effect." Motor vehicle accident mortality dropped during the study period, perhaps reflecting efforts to control the problem. Suicide rates were depressed during the period of active duty. There was a suggestion that cancer mortality was associated with submarine type; however, the age distribution casts doubt that the excess was occupationally induced.