Electromyography (EMG) is a technique that registers muscle activity allowing the study of muscle behavior. One such study is the muscle fatigue analysis, which can be defined as the beginning of muscle tiredness and can occur in static or dynamic contractions. There are moments of activity and muscle rest during dynamic contractions, where the latter may not be interesting for some muscle analysis, like muscle fatigue. Therefore, removing muscle inactivity is attractive, but as the signal is not standardized, this process is slower and more individualized. So, this work aims at presenting a muscle fatigue analysis using semi-automatic segmentation of the EMG signal, cutting off the inactivity moments of an electromyography signal obtained during dynamic activity. This analysis consists in calculating the median frequency, a traditional measure to analyze muscle fatigue in static contractions. To assess the efficiency of the proposed method, the median frequency values were compared using manual and semi-automatic segmentation. The result is satisfactory, preserving the desired signal parts for analysis and indicating muscle fatigue as expected.