At the first synapse of the vertebrate visual pathway, photoreceptors perform the extraordinary task of releasing neurotransmitter via exocytosis in a graded manner for extended periods of time. Although the release of neurotransmitter via exocytosis is a highly conserved, fundamental feature of nervous system function, the extent and pattern of neurotransmitter release can be shaped by a variety of intrinsic and extrinsic factors. To gain a better understanding of the intrinsic properties that shape synaptic output and contribute to the ability of photoreceptors to release neurotransmitter in a tonic and graded manner, we performed time-resolved measurements of synaptic vesicle dynamics in isolated vertebrate photoreceptors. Photoreceptors were acutely dissociated from the retina of aquatic tiger salamanders. Changes in membrane surface area indicative of synaptic vesicle fusion and the exocytotic release of neurotransmitter were detected using membrane capacitance measurements. Calcium currents were measured using standard whole-cell patch clamp techniques, and intracellular calcium was measured using ratiometric fluorescent calcium indicator dyes. Exocytosis was evoked either by membrane depolarization and the activation of calcium entry through voltage-gated channels or via the rapid and global release of intraterminal calcium by the flash-photolysis of caged calcium. When exocytosis was stimulated via membrane depolarization in cone photoreceptors, two kinetic components of release were revealed. The first component was discrete in size, reaching a steady-state value of ≈ 45 fF, corresponding to approximately 1,000 vesicles. Fusion of vesicles in this pool was blocked by low millimolar EGTA. This pool could be depleted with a time constant of a few hundred milliseconds. Thus, the first pool has features similar to the releasable pool of synaptic vesicles described in other neurons. However, unlike other releasable pools, the cone releasable pool recovered from depletion quite rapidly (t ≤ 1 s). Endocytosis was sufficiently slow that it is unlikely that refilling of the releasable pool occurred via newly-retrieved vesicles. The second component of release was approximately twice the size of the releasable pool and had a time constant for depletion that was substantially slower (t ≈ 3 s). Simulations using a computational model of release demonstrated that the rapid replenishment of the releasable pool from the secondary pool was sufficient to maintain release evoked by near-maximal stimulations of up to 5 seconds in duration. The inclusion of a tertiary pool of vesicles, corresponding to the remaining cytoplasmic pool of vesicles, allowed for maintained release in the face of even longer periods of near-maximal stimulation. The present data, combined with our earlier findings that the relationship between exocytosis and calcium is unusually shallow in photoreceptors, can account for several of the unique features of exocytosis in these neurons. First, in the range of intraterminal calcium believed to be physiologically-relevant, the rate of exocytosis almost linearly tracks the local calcium concentration, contributing to the graded nature of transmitter release. Secondly, a relatively large releasable pool with a fast rate of refilling from preformed vesicles contributes to the ability to maintain release for extended periods.