1. The glucose and ATP dependence of exocytosis were investigated in single mouse pancreatic B-cells by monitoring changes in cell capacitance evoked by voltage-clamp depolarizations, infusion of high [Ca2+]i buffers or photorelease of caged Ca2+ or ATP. 2. In intact B-cells, using the perforated patch whole-cell technique, glucose (5 mM) increased exocytotic responses evoked by membrane depolarization 5-fold over that observed in the absence of the sugar. Increasing the glucose concentration to 20 mM produced a further doubling of exocytosis. The stimulatory action of glucose was attributable to glucose metabolism and abolished by mannoheptulose, an inhibitor of glucose phosphorylation. 3. Exocytosis triggered by infusion of high [Ca2+]i and ATP was reduced by 80% when ATP was replaced by its non-hydrolysable analogue adenosine 5'-[beta, gamma-methylene]triphosphate (AMP-PCP) in standard whole-cell experiments. Exocytosis elicited by GTP gamma S was similarly affected by replacement of ATP with the stable analogue. 4. Photoreleasing ATP in the presence of 170 nM [Ca2+]i, following the complete wash-out of endogenous ATP produced a prompt (latency, < 400 ms) and biphasic stimulation of exocytosis. 5. Elevation of [Ca2+]i to exocytotic levels by photorelease from Ca(2+)-nitrophenyl EGTA preloaded into the cell evoked a biphasic stimulation in the presence of Mg-ATP. The response consisted of an initial rapid (completed in < 200 ms) phase followed by a slower (lasting > or = 10 s) sustained component. Replacement of ATP with AMP-PCP abolished the late component but did not affect the initial phase. The latency between elevation of [Ca2+]i and exocytosis was determined as < 45 ms. Inclusion of N-ethylmaleimide (NEM; 0.5 mM for 3 min) in the intracellular solution exerted effects similar to those obtained by substituting AMP-PCP for ATP. 6. We conclude that the B-cell contains a small pool (40 granules) of primed granules which are immediately available for release and which are capable of undergoing exocytosis in an ATP-independent fashion. We propose that this pool of granules is preferentially released during first phase glucose-stimulated insulin secretion. The short latency between the application of ATP and the onset of exocytosis finally suggests that priming takes place with sufficient speed to participate in the rapid adjustment of the secretory capacity of the B-cell.