We studied the effect of an increase in physical activity on energy balance and body composition without interfering with energy intake (EI). Sixteen women and sixteen men, aged 28-41 years, body mass index 19.4-26.4 kg/m2, not participating in any sport before the start of the experiment, prepared to run a half-marathon competition after 44 weeks. Measurements of body composition, EI and energy expenditure (EE) were performed before (0 weeks), and 8, 20, and 40 weeks after the start of training. Body composition was measured with hydrodensitometry and isotope dilution, and EI with a 7 d dietary record. EE was measured overnight in a respiration chamber (sleeping metabolic rate (SMR)) and in a number of subjects over 2-week intervals with doubly-labelled water (average daily metabolic rate (ADMR)). ADMR showed an average increase of 30% in both sexes from the start of training onwards while SMR tended to decrease. EI showed a tendency to drop from week 20 to week 40 in the men and a tendency to increase from week 20 to week 40 in the women. Body mass (BM) did not change in both sexes until the observation at 40 weeks when the median value of the change in men was -1.0 kg (P < 0.01; Wilcoxon signed-rank) while the corresponding change of -0.9 kg in the women was not statistically significant. Body composition changes were most pronounced in men as well. Based on changes in BM, body volume and total body water, men lost 3.8 kg fat mass (FM) (P < 0.001; Wilcoxon signed-rank) and gained 1.6 kg protein mass (P < 0.01; Wilcoxon signed-rank) while the corresponding changes in women were 2.0 kg (P < 0.05; Wilcoxon signed-rank) and 1.2 kg (P < 0.05; Wilcoxon signed-rank). In men the loss of FM was positively correlated with the initial percentage body fat (Pearson r 0.92, P < 0.001). In conclusion, body fat can be reduced by physical activity although women tend to compensate for the increased EE with an increased EI, resulting in a smaller effect on BM and FM compared with men.