In this work we propose a novel method based on sEMG signals, easy and fast to perform, administered with a robotic device to maximize repeatability and objectivity. Muscle fatigue, which is frequently experienced by healthy subjects, can be a highly debilitating symptom in case of neuromuscular disorders. Its assessment provides crucial information on the progression of the disability itself, on patient's muscular function and on the efficacy of the eventual clinical intervention. Hence, a robust and objective protocol for fatigue assessment is fundamental in rehabilitation practice. Therefore, the aim of this work was twofold. Firstly, we aimed to test the proposed method and highlight its strengths and drawbacks for a future optimization and implementation in a clinical context. Secondly, we meant to identify which are the most sensitive and reliable measures of muscles' performance that can quickly and optimally predict subjects' behavior. sEMG signals were collected from right Extensor and Flexor Carpi Radialis of 9 healthy subjects during a flexion-extension robotic task consisting in a haptic tracking in a viscous field. Three indicators of fatigue (Mean Frequency, Dimitrov Index, Root Mean Square) were obtained and we proposed a novel sensitive parameter which determines the Onset of Fatigue.