The role of ryanodine-sensitive intracellular Ca2+ stores present in nonmuscular cells is not yet completely understood. Here we examine the physiological parameters determining the dynamics of caffeine-induced Ca2+ release in individual fura 2-loaded sympathetic neurons. Two ryanodine-sensitive release components were distinguished: an early, transient release (TR) and a delayed, persistent release (PR). The TR components shows refractoriness, depends on the filling status of the store, and requires caffeine concentrations > or = 10 mM. Furthermore, it is selectively suppressed by tetracaine and intracellular BAPTA, which interfere with Ca(2+)-mediated feedback loops, suggesting that it constitutes a Ca(2+)-induced Ca(2+)-release phenomenon. The dynamics of release is markedly affected when Sr2+ substitutes for Ca2+, indicating that Sr2+ release may operate with lower feedback gain than Ca2+ release. Our data indicate that when the initial release occurs at an adequately fast rate, Ca2+ triggers further release, producing a regenerative response, which is interrupted by depletion of releasable Ca2+ and Ca(2+)-dependent inactivation. A compartmentalized linear diffusion model can reproduce caffeine responses: When the Ca2+ reservoir is full, the rapid initial Ca2+ rise determines a faster occupation of the ryanodine receptor Ca2+ activation site giving rise to a regenerative release. With the store only partially loaded, the slower initial Ca2+ rise allows the inactivating site of the release channel to become occupied nearly as quickly as the activating site, thereby suppressing the initial fast release. The PR component is less dependent on the store's Ca2+ content. This study suggests that transmembrane Ca2+ influx in rat sympathetic neurons does not evoke widespread amplification by CICR because of its inability to raise [Ca2+] near the Ca2+ release channels sufficiently fast to overcome their Ca(2+)-dependent inactivation. Conversely, caffeine-induced Ca2+ release can undergo considerable amplification especially when Ca2+ stores are full. We propose that the primary function of ryanodine-sensitive stores in neurons and perhaps in other nonmuscular cells, is to emphasize subcellular Ca2+ gradients resulting from agonist-induced intracellular release. The amplification gain is dependent both on the agonist concentration and on the filling status of intracellular Ca2+ stores.