The kinetics of Ca2(+)-dependent release of glutamate from guinea-pig cerebrocortical synaptosomes evoked by KCl or 4-aminopyridine are investigated using a continuous fluorimetric assay. Release by both agents is biphasic, with a rapid phase complete within 2 s followed by a more extensive slow phase with a half-maximal release in 52 s for KCl-evoked release and greater than 120 s for 4-aminopyridine-evoked release. The two phases of glutamate release may reflect a dual localization of releasable vesicles at the active zone and in the bulk cytoplasm. Decreasing depolarization depresses the extent rather than increasing the time for half-maximal Ca2(+)-dependent release. Both the fast and the slow phases of glutamate release require external Ca2+ and cytoplasmic ATP. KCl depolarization produces a transient "spike" of cytoplasmic free Ca2+ [( Ca2+]c), which recovers to a plateau; the major component of glutamate release occurs during this plateau. Predepolarization in the absence of added external Ca2+, to inhibit transient Ca2+ channels, does not affect the subsequent glutamate release evoked by Ca2+ readdition. Thus, release involves primarily noninactivating Ca2+ channels. For a given increase in [Ca2+]c, KCl and 4-aminopyridine cause equal release of glutamate, while ionomycin releases much less glutamate. This lowered efficiency is not due to ATP depletion. It is concluded that glutamate exocytosis is evoked by localized Ca2+ entering through noninactivating voltage-dependent Ca2+ channels and that nonlocalized Ca2+ entry with ionomycin is inefficient.