The most commonly used aerobic training models for experimental research on animals are running and swimming exercises with moderate intensities and with fixed volumes. In models that mimic human menopause, the omission or simple contempt of body composition and fat redistribution can falsify results in terms of performance and training effects. Therefore, this study aimed to clarify the effect of body composition on the performance of rats submitted to swimming and running training. We used 24 twelve-week old virgin, female Wistar rats and divided them into 4 groups: ovariectomised swimmers (OV-ST) and runners (OV-RT); and sham surgery swimmers (SH-ST) and runners (SH-RT). All animals performed 4 weeks of training in accordance with their respective group. After the experimental period, the ovariectomised animals exhibited an increased relative weight and fat mass compared to SH groups, but their gain of performance data were different according to the kind of training they received; the OV-ST group had a higher performance compared to the OV-RT and SH-RT groups. It is likely that the higher body weight gain of the OV-ST group animals is a consequence of training with lighter loads, resulting from the higher buoyancy in animals with greater adiposity, reducing the effect of exercise on the control of body composition and weight. Accordingly, we conclude that the swimming training, as well as the maximum stress tests in the water, do not seem to be the best training option in animals prone to obesity. We show that body composition influences the results of animal performance and workload in water. Thus, body composition must be considered and overload methods in water must be adjusted to achieve desired training results.