A number of calcium buffers were examined for their ability to reduce evoked transmitter release when injected into the presynaptic terminal of the squid giant synapse. Injection of EGTA was virtually ineffective at reducing transmitter release, even at estimated intracellular concentrations up to 80 mM. Conversely, the buffer 1,2-bis(2-aminophenoxy)ethane-N,N,N',N'-tetraacetic acid (BAPTA), which has an equilibrium affinity for calcium similar to that of EGTA at pH 7.2, produced a substantial reduction in transmitter release when injected presynaptically. This effect of BAPTA was reversible, presumably because the buffer diffused out of the terminal and into uninjected regions of the presynaptic axon. BAPTA derivatives with estimated intracellular calcium dissociation constants (Kd) ranging from 0.18 to 4.9 microM were effective at reducing transmitter release at similar estimated concentrations. A BAPTA derivative with an estimated intracellular Kd of 31 mM was less effective. BAPTA did not affect presynaptic action potentials or calcium spikes in ways that could explain its ability to reduce transmitter release. The relative effects of presynaptic injections of BAPTA and derivatives are consistent with the calcium-buffering capability of these compounds if the presynaptic calcium transient that triggers release is hundreds of microM or larger. The superior potency of BAPTA compared to EGTA apparently results from the faster calcium-binding kinetics of BAPTA and suggests that the calcium-binding molecule that triggers release binds calcium in considerably less than 200 microsec and is located very close to calcium channels.