A retrospective longitudinal analysis of fat distribution and cause-specific mortality was performed on data from 105,062 men discharged from the United States Army in 1946-47. Baseline height, weight, waist and hip girth, and 23-year follow-up mortality data were available for 84,910 white men. Proportional hazard survival analysis was performed by 5-year age group for waist/hip ratio (WHR) and for body mass index (BMI) in prediction of time to death from ischaemic heart disease (IHD) and stroke. Relative risk of IHD fatality per standard deviation (s.d.) of WHR ranged from 1.11 to 1.17, the higher values appearing in younger age groups. Relative risk due to BMI was not significant in the group aged 16-20 years at time of discharge from service, but ranged from 1.22 to 1.25 per s.d. among the 21-30 year olds. WHR was predictive of cerebrovascular disease mortality among 16-25 year olds, carrying a relative risk of 1.24 to 1.35 per s.d. BMI was not predictive of cerebrovascular disease mortality in any age group. Multivariate models indicated that WHR and BMI were related to subsequent IHD independently of each other in most age groups. WHR and BMI both contribute to risk of premature IHD mortality and WHR to risk of cerebrovascular disease mortality in an initially relatively healthy population of young men, although the effects are not equivalent in all age groups.