The study presents an analytical, simulation, and experimental analysis of amplitude cancellation of motor unit action potentials in the interference EMG and its relation to the size of the spike-triggered average (STA) EMG. The amount of cancellation of motor unit action potentials decreases monotonically as a function of the ratio between the root mean square (RMS) of the motor unit action potential and the RMS of the interference EMG signal. The theoretical derivation of this association indicates a method to measure cancellation in individual motor units by STA of the interference and squared EMGs. The theoretical relation was examined in both simulated EMG signals generated by populations of 200 motor units and experimental recordings of 492 and 184 motor unit action potentials in the vastus medialis and abductor digiti minimi muscles, respectively. Although the theoretical relation predicted (R² = 0.95; P < 0.001) the amount of cancellation in the simulated EMGs, the presence of motor unit synchronization decreased the strength of the association for small action potentials. The decrease in size of the STA obtained from the squared EMG relative to that extracted from the interference EMG was predicted by the experimental measure of cancellation (R² = 0.65; P < 0.001, for vastus medialis; R² = 0.26; P < 0.05, for abductor digiti minimi). The results indicate that cancellation of action potentials in the interference EMG can be analytically predicted and experimentally measured with STA from the discharge times of the motor units into the surface EMG.