The discovery that Ca2+ is necessary for the release of neurotransmitter, the primary means by which nerve cells communicate, led to the calcium hypothesis of neutransmitter release, in which release is initiated after an action potential only by an increase in intracellular Ca2+ concentration near the release sites and is terminated (1-2 ms) by the rapid removal of Ca2+. Since then, the calcium-voltage hypothesis has been proposed, in which the depolarization of the presynaptic terminals has two functions. First, in common with the calcium hypothesis, the Ca2+ conductance is increased, thereby permitting Ca2+ entry. Second, a conformational change is induced in a membrane molecule that renders it sensitive to Ca2+, and then binding of Ca2+ to this active form triggers release of neurotransmitter. When the membrane is repolarized, the molecule is inactivated and release is terminated, regardless of the local Ca2+ concentration at that moment. This hypothesis, in contrast to the calcium hypothesis, accounts for the insensitivity of the time course of release to experimental manipulations of intracellular Ca2+ concentration. Furthermore, it explains rapid termination of release after depolarization, even though Ca2+ concentration may still be high. Here we describe experiments that distinguish between these two hypotheses and find that our results support the calcium voltage hypothesis.