A longitudinal study was conducted on 398 athletes and 369 nonathletes who were born before 1920 and attended Michigan State University. The subjects were compared to determine if intercollegiate athletic competition accounts for significant variation in longevity when considered with somatotype. Because some of the subjects were still alive at the time of the study, the BMDP Statistical Software was used to do a survival analysis with covariates. Preliminary comparisons considered the differences in somatotype between athletes and nonathletes. Two sample t-tests indicated that athletes were more mesomorphic and less ectomorphic (p less than .05) than nonathletes. When comparing the relationship between somatotype and longevity, the pooled data of athletes and nonathletes indicated that endomorphs were shorter lived than the other three comparison groups. When only the athletes were considered, similar results were found. However, the nonathlete group exhibited differences only between the mesomorphic and endomorphic groups. The endomorphs were shorter lived. Longevity was examined by using the Cox proportional hazards regression method with somatotype and athlete/nonathlete status as covariates. Somatotype, by itself, was found to be significantly related to longevity, (p less than .001). Athletic status was not significantly related to longevity, either by itself or when entered into the model with somatotype.