Aim The well-documented association of reductions in physical activity with age is based largely on cross-sectional studies that do not distinguish between prevalence of physical activity and initiation/discontinuation patterns. These studies also fail to consider the changes in physical activity as a function of life stage. We investigated how physical activity evolves over a lifetime. We distinguished between prevalence and initiation/discontinuation patterns and also considered multidirectional age effects. Subject and methods The analyses are based on the annual German Socio-Economic Panel survey, which originally involved 6,000 households (>12,000 individuals) from age 16. Our study included 3,487 subjects (West Germany), i.e., 33% of the individuals first interviewed in 1984. A survival analysis model was constructed to assess the simultaneous effect of age and generation, while adjusting for potential confounders.Results The results show that the ubiquitous differences in physical activity between age groups are mainly attributable to cohort differences, not age effects. The likelihood of initiating at least weekly physical activity declines with age, as does the likelihood of discontinuing an established exercise habit. Both trends are more readily explained by changes in occupational and family biography than by health and fitness. Conclusion The ubiquitous differences in physical activity between age groups are largely due to intergenerational differences.