We explore the properties of models of synaptic vesicle dynamics, in which synaptic depression is attributed to depletion of a pool of release-ready vesicles. Two alternative formulations of the model allow for either recruitment of vesicles from an unlimited reserve pool (vesicle state model) or for recovery of a fixed number of release sites to a release-ready state (release-site model). It is assumed that, following transmitter release, the recovery of the release-ready pool of vesicles is regulated by the intracellular free Ca(++) concentration, [Ca(++)](i). Considering the kinetics of [Ca(++)](i) after single presynaptic action potentials, we show that pool recovery can be described by two distinct kinetic components. With such a model, complex kinetic and steady-state properties of synaptic depression as found in several types of synapses can be accurately described. However, the specific assumption that enhanced recovery is proportional to [Ca(++)](i), as measured with Ca(++) indicator dyes, is not confirmed by experiments at the calyx of Held, in which [Ca(++)](i)-homeostasis was altered by adding low concentrations of the exogenous Ca(++) buffer, fura-2, to the presynaptic terminal. We conclude that synaptic depression at the calyx of Held is governed by localized, near membrane [Ca(++)](i) signals not visible to the indicator dye, or else by an altogether different mechanism. We demonstrate that, in models in which a Ca(++)-dependent process is linearly related to [Ca(++)](i), the addition of buffers has only transient but not steady-state consequences.