Ten pairs of monozygotic twins of both sexes were submitted to a 20-wk endurance-training program, four and five times per week, 40 min per session, at an average of 80% of the maximal heart rate reserve. Testing and training were performed on cycle ergometers. Maximal aerobic power (MAP in ml O2 X min-1 X kg-1) and ventilatory aerobic (VAT) and anaerobic (VANT) thresholds (ml O2 X min-1 X kg-1) were measured before and after the training program, as well as during the 7th and 14th week to adjust training to changes in maximal heart rate. Considering the 20 individuals as a group, training significantly (P less than or equal to 0.01) increased MAP (from 44 +/- 6 to 50 +/- 6), VAT (25 +/- 3 to 30 +/- 4), and VANT (36 +/- 5 to 42 +/- 6). Thus, MAP improved by 12% of the pre-test value, while mean changes in VAT and VANT reached 20% and 17%, respectively. There were, however, considerable interindividual differences in training gains as exemplified by a range of about 0% to 41% for MAP. Differences in the MAP response to training were not distributed randomly among the twin pairs. Thus, intraclass correlations computed with the amount of improvement in MAP (ml O2 X min-1 X kg-1) reached 0.74 (P less than 0.01) indicating that members of the same twin-pair yielded approximately the same response to training. The same coefficient reached 0.43 and 0.24 for VAT and VANT, respectively (P greater than 0.05). These results suggest that there are considerable individual differences in the adaptive capacity to short-term endurance training. Moreover, sensitivity of maximal aerobic power to such training is largely genotype-dependent.