Presynaptic and post-synaptic cells of the squid giant synapse were voltage-clamped simultaneously to study the relationship between presynaptic Ca current and transmitter-induced post-synaptic current (p.s.c.). Local Ca application was used to restrict Ca current and transmitter release to a limited region of the presynaptic terminal and thus minimize errors due to spatial heterogeneity of presynaptic membrane potential. Presynaptic terminals were depolarized by brief (3-6 ms) voltage-clamp pulses of varying amplitude to collect graded series of presynaptic Ca current and p.s.c. records. During presynaptic depolarization at 14 degrees C, Ca current activation preceded initial onset of p.s.c. (on-p.s.c.) by an interval of approximately 1 ms. The main component of on-p.s.c. followed Ca current activation by about 2 ms. The delay between a brief Ca tail current and peak response of the p.s.c. produced after pulse termination (off-p.s.c.) was also approximately 2 ms. Curves relating both Ca current and p.s.c. magnitudes to presynaptic potential were bell shaped with peaks near -10 mV, but the p.s.c. curve showed stronger voltage dependence on both sides of the peak. With very small and very large presynaptic command pulses, Ca current could be observed without measureable p.s.c. Synaptic transfer curves, plotting p.s.c. as a function of presynaptic Ca current, resembled third-power functions. On the average, p.s.c.s fit a curve representing the 2.9 power of Ca current (range 2.4-3.5 in eighteen experiments). Transfer curves consisted of two limbs: one from presynaptic pulses below -10 mV and the other from more positive pulses. These two limbs were similar and generally resembled power functions of identical exponent. It is thus likely that the third-power function accurately reflects synaptic current transfer, rather than interference from some other voltage-dependent process. Power functions fitting small-pulse and large-pulse limbs of some transfer curves had different scale coefficients, even though exponent values were the same. Consideration of synaptic transmission kinetics suggests that the voltage dependence of Ca channel opening rates can probably explain the difference in transfer curve limbs. Our experiments provide no evidence for an intrinsic voltage dependence of the transmitter release process.