Exercise is considered medicine; however, the individual degree of responsiveness to a standardized dose of exercise is idiosyncratic. Individual responsiveness between distinct exercise modalities and the genetic/environmental contributions to exercise response are not well understood. In this novel randomized cross-over design study, monozygotic and dizygotic twins, as pairs, underwent 3 months of resistance and endurance training, separated by a 3 month washout period, aiming to assess training responses in strength and fitness outcomes to dichotomous training modalities, as well as the genetic/environmental contributions to exercise response. Our findings indicate that (i) individual responsiveness differs between exercise modalities; (ii) low-responders to one mode may be 'rescued' by switching to an alternate mode of exercise; and (iii) genes may not play as large a role, as previously estimated from cross-sectional data, for exercise training adaptation. The present study has implications for those charged with optimizing the benefits of exercise by means of individualizing the exercise prescription. Exercise response is idiosyncratic, although the degree of responsiveness, concordance in response between modalities and genetic contribution to responsiveness are not well understood. We investigated this using a novel randomized cross-over design of dichotomous exercise interventions in mono-(MZ) and di-zygotic (DZ) twin pairs. We studied strength (1RM) and fitness (V?O2max) responses in 84 same-sex untrained twins (30 MZ, 12 DZ pairs; 24.9 +/- 5.4 years). Twins, as pairs, underwent 3 months of resistance (RES) and endurance (END) training, separated by a 3 month washout period. Training responses and genetic/environmental contributions to responses were assessed. Leg strength 1RM increased following RES but not END (o47 +/- 29vs. 3 +/- 26 kg;P < 0.001), whereasV?O2maxincreased following END but not RES (o0.25 +/- 0.26vs. 0.04 +/- 0.25 L min(-1);P < 0.001). A higher percentage of individuals responded to RES for strength and to END forV?O2max(P < 0.0001). Within-individual responses to each mode were not correlated (P > 0.05). Cross-sectional intraclass correlations were higher for MZ than DZ pairs for all variables, largely as a result of shared environment. Following training, MZ, but not DZ pairs, were significantly correlated for strength change to RES (r(MZ) = 0.62,P = 0.002) and END (r(MZ) = 0.36,P = 0.04), and forV?O2maxchange to END (L min(-1),r(MZ) = 0.45,P = 0.02) with a mixture of unshared/shared environmental contributions. Our findings indicate that individual responsiveness differs between modalities and low-responders to one mode may be 'rescued' by switching to an alternate mode. Additionally, genes may not play as large a role as previously estimated from cross-sectional data for training adaptation, and/or cross-sectional data do not reflect longitudinal training effects. The present study has implications for optimizing the individualization of exercise prescription.