Abstract Introduction The duration of bursts of muscle activity e.g. while running or walking, is too short for long pulse trains of motor unit action potentials (MUAPs) to develop. A pool of motor units is likely activated simultaneously which generates clustered MUAPs. The hypothesis is that the EMG power spectra are modulated by the fact that MUAPs cluster. The purpose is to quantify this modulation analytically. Methods A model of an EMG signal is presented that includes clustered MUAPs. Results According to the model the influence of MUAPs clustering is shown to be largest at lower frequencies and increases when the width of the time window containing the clusters decreases. Discussion The power at frequencies below 60 Hz strongly reflects changes of the degree of clustering. The mean frequency of the EMG therefore decreases when MUAPs cluster more tightly. Thus, clustering of MUAPs competes with other physiological properties that influence the mean frequency. The EMG power is proportional to the number of active MUAPs at high frequencies but approaches a value proportional to the square of the number of active MUAPs at very low frequencies. To obtain a measure of amplitude that is proportional to the number of active motor units, one should focus on the higher frequency power components only, however, to monitor the effect of clustering of MUAP one should focus on the lower frequency power. That could become relevant for comparing pre- and post-operative clinical gait studies where changes in MUAPs clustering may play a significant role.