One of the continuing paradoxes facing social epidemiologists concerns sex differences in morbidity and mortality. Although women live longer than men, they apparently get sick more. We hypothesize that women's higher morbidity levels result from less paid work and lower wages combined with more hours spent in household labor, child care, and helping others, and fewer hours of leisure and sleep. Men and women hold different social roles; men hold most of the highly rewarding roles. We operationalize social roles as time commitments to various role-related activities. This approach provides interval-level measures such as time spent in caring for children instead of simple dichotomies such as parent/nonparent. We find that when gender differences in social roles are controlled, being male is associated with poorer health than being female. We conclude that if gender roles were more equal, women would experience better health than men, more consistent with their greater longevity.