We studied the effects of varying extracellular Ca2+ ([Ca2+]o) and Ca2+ channel density and intracellular loading of Ca2+ chelators on stimulation-induced rises in intracellular Ca2+ ([Ca2+]i) in frog motor nerve terminals with Ca2+ imaging. The slowly waxing and waning components of rises in [Ca2+]i induced by repetitive tetani were suppressed by blockers of Ca2+ pumps of the endoplasmic reticulum (thapsigargin and cyclopiazonic acid) and a blocker of ryanodine receptors [8-(N,N-diethylamino)octyl 3,4,5-trimethoxybenzoate hydrochloride] without affecting the initial quickly-rising component, thus reflecting the priming (and then subsequent rapid activation) and inactivation phases of Ca2+-induced Ca2+ release (CICR) from the endoplasmic reticulum. A short tetanus-induced rise in [Ca2+]i was proportional to [Ca2+]o, whereas the component of CICR was non-linearly related to [Ca2+]o with saturation at 0.9 mm. The progressive blockade of Ca2+ channels by ω-conotoxin GVIA caused proportional decreases in CICR and short tetanus-induced [Ca2+]i rises. Intracellular loading of BAPTA and EGTA reduced the magnitude of CICR as well as short tetanus-induced rises in [Ca2+]i with a greater effect of BAPTA than EGTA on CICR. The time to peak and the half decay time of CICR were prolonged by a low [Ca2+]o or Ca2+ channel blocker or [Ca2+]i chelators. These results suggest that ryanodine receptors sense the high [Ca2+]i transient following single action potentials for triggering CICR, whereas the priming and inactivation processes of CICR sense a slower, persisting rise in [Ca2+]i during and after action potential trains. A model is presented that includes CICR activation in elementary units.