Previous experiments have indicated that there is a continuous leakage of acetylcholine (ACh) from resting motor nerve terminals which can produce a small depolarization in anti-esterase treated endplates (Katz & Miledi 1977; Vyskocil & Illes 1978). This leakage might be expected to be intensified during the presynaptic action potential and so lead to a very small non-quantal endplate response. This hypothesis was examined, in frog and mammalian endplates, by stimulating the motor nerve in a calcium-deprived medium and recording the summated average response to several hundred stimuli. The result was completely negative; no trace of a non-quantal endplate potential was ever observed, with the limit of detection being always less than 10 $\mu $V, and sometimes as low as 2 $\mu $V. These experiments suggest that the leakage of ACh either does not originate predominantly from the synaptic region of the axon terminal, or that it occurs by a mechanism that is not directly influenced by the membrane potential.