In the field of physical activity a significant importance is held by aerobic endurance training, which is a relatively low intensity exercise that depends primarily on aerobic energy generating processes. This type of training is used for the overall endurance and fitness of the body by engaging all the major systems of the body and pushing the limits of their functions with the ultimate goal of adapting these structures to the stress of the physical effort and thus improving the performance output. However, using classic training methods, it is often impossible to ensure the necessary operating points in order to obtain the desired results. The human heart rate model with respect to the velocity of walking or running presents inherent nonlinearities that are required to be taken into account when designing computer integrated training aids. This paper implements such a tool using a fuzzy control system for human heart rate during aerobic endurance training. The controller is tested and validated using two nonlinear human-on-treadmill models, and it presents robustness to significant model variations, as well as uncertainties in the model parameters and white noise type of disturbances. The designed control system ensures desired heart rate profiles, finding its usefulness in supporting the training configuration process by specialized professionals.