This study investigates the relative extents to which health-related selection and the occupation itself might explain increased mortality and morbidity among seafarers. The study group comprised a cohort of young men conscripted for compulsory military training in Sweden during 1969-1970. At conscription, information on ill health and also a number of health-related factors, such as alcohol consumption, was collected. Information on occupation over the period 1970-1990 was extracted from census data. Subjects were 206 men who had been seamen in either 1970 or 1975, or both. These men were compared with 16,292 other men who were unskilled workers during these years. Information on mortality, psychiatric diagnoses and early retirement was collected from relevant registers. Health-related selection was found among persons entering the occupation of seaman. Seamen were also found to have increased relative risks of mortality, early retirement, and psychiatric diagnosis (both related to and not related to alcohol or drug abuse), even after taking into account several background variables concerned with circumstances during childhood and adolescence. Health-related selection was only partly able to explain the increased relative risks faced by seamen found in this study. The occupation itself remains a strong risk indicator, even after controlling for a large number of selection factors.