Phasic release from motor-nerve terminals of crayfish and frogs was elicited and recorded by means of a macro-patch-clamp electrode through which the terminal was depolarized in graded pulses. The tip of the electrode was perfused and the Ca concentration around the terminal, Cae, was controlled independent from that in the superfusion of the muscle, Cab. Release increased with pulse duration with a double-logarithmic slope of 5 to 9 in crayfish and frogs, which represents a form of "early facilitation" (Katz and Miledi 1968). In crayfish, this relation was shifted to longer pulse durations on lowering Cae, while in frogs, in addition, the saturation level of release was suppressed at low Cae. Responses to twin pulses with intervals of 7-10 ms showed facilitation, Fd. When pulse duration of the twin pulses was increased, starting from about 0.5 ms, Fd increased to a maximum, but declined for longer pulses which elicited release approaching the saturation range. On lowering Cae, the maximum of Fd, Fd, increased in amplitude and was shifted to larger pulse durations. Also reduction of Cab increased Fd. The effects of pulse duration and of Cae and Cab on Fd are predicted by the residual Ca theory of facilitation, if it is assumed that changes of Cae produce corresponding changes in Ca inflow during depolarization, and if the resting intracellular Ca concentration is influenced by the extracellular Ca concentration. The large values of early facilitation can not be explained by the residual Ca theory of facilitation and may indicate the action of another depolarization dependent factor which joins in the control of release.