1. Using the theory of noise analysis an attempt was made to measure frequency and amplitude of miniature end-plate potentials (MEPPs) under conditions of vigorous transmitter release. Frog sartorius muscles were incubated in a depolarizing (32 mM-K+) medium which lacked Ca2+ to prevent transmitter release. Subsequently, when the membrane potential had become stable at about -40 mV, end-plates were superfused with 4 mM-Ca2+-containing medium for 1 min periods with 5 min intervals between the superfusions. 2. Most junctions ('fast' type) responded to Ca2+ with a relatively large, noisy depolarization (5.8-14.5 mV) which subsided rapidly during subsequent challenges with Ca2+. Other junctions ('slow' type) responded with only 1-1.6 mV depolarizations which were rather well sustained during the consecutive Ca2+ applications. 3. From the variance, E2, and the depolarization, V, caused by Ca2+ the frequency n and amplitude factor q of the MEPPs were calculated. Values of n were 3-4 x 10(4) and 0.1-1 x 10(4) s-1 in the fast- and slow-type junctions, respectively. The mean value of q was 0.16 mV; it remained more or less constant in the fast-type junctions, but tended to decline in the slow-type junctions. 4. As expected, cholinesterase inhibitors potentiated V and E2 as well as individual MEPPs. However, no advantage could be taken from this finding, since these drugs caused burst-like peaks superimposed on the voltage signal, precluding application of noise analysis. 5. The results strongly suggest that, at least in the fast-type junctions, K+ caused an extremely rapid depletion of the store of transmitter quanta, whose mean size did not change appreciably in the course of the experiment. However, in the slow-type junctions during prolonged incubation, it cannot be excluded that the gradual decline of q was due to the release of newly formed, unripe quanta.