An important measure of the performance of a myoelectric control channel for powered artificial limbs is the myoelectric signal processor output signal-to-noise ratio (SNR). The signal and noise in this context are, respectively, the mean and variance of the estimate of some signal parameter to be used for control purposes. These quantities are determined by the signal processor, motor unit recruitment and motor unit firing statistics. The paper investigates, through analytical, simulation and experimental work, the role and significance of recruitment and firing statistics in channel performance. Equations are derived which express, for the single and multi-unit cases, channel SNR as a function of the number of active units, firing rates, action potential amplitude variation and action potential moments. A computer-simulated myoelectric signal is generated in which these variables can be controlled and SNR measured. The simulation results are compared with the theoretical and found to agree very well. Limited experiments with wire intramuscular electrodes and surface electrodes are performed to measure in vivo SNR from the biceps brachii muscle. The results of the experiments agree well with those of the simulation and theoretical work. The significance of this work is that it provides insight into the roles of important physiological parameters in myoelectric channel performance. It will also provide data necessary for the development of SNR enhancement techniques.