We are lacking very long-term and controlled intervention studies investigating the effects of habitual physical activity on health-related factors. To address this gap, we performed a natural experiment by identifying same-sex twin pairs in which the co-twins of each pair differed with respect to leisure-time physical-activity habits throughout their adult life. Our criterion for the discordance was that the same co-twin had a higher leisure time-activity volume than that of the other member of the pair at the majority -- if not all -- of the follow-up time points according to reported/interviewed physical-activity data. Overall, we identified and conducted multidimensional health-related measurements (including fitness, body composition, cardiometabolic risk factor levels, bone and arterial status, and exercise motivation) of 16 twin pairs (seven monozygotic (MZ) and nine dizygotic (DZ) pairs, mean age 60 years) who had persistent discordance in leisure-time physical-activity habits over three decades (TWINACTIVE study). In our discordant-pair study design, after adjusting for sequence-level genes, both systemic-level metabolic, and site-specific structural findings differed significantly in the pairwise analysis in MZ pairs only. These findings included intrapair differences in accumulated fat depots and structure of heart, arteries, and bones. In addition, our study revealed intrapair differences in metabolic and regulatory pathways, which may partly explain the mechanistic links between long-term physical activity, phenotypic changes, and decreased risk of cardiometabolic diseases.