1. When pancreatic islets are preincubated for 20h in the presence of glucose (83.3mM) and thereafter transferred to a glucose-free medium, theophylline (1.4mM) provokes a dramatic stimulation of insulin release. This phenomenon does not occur when the islets are preincubated for either 20h at low glucose concentration (5.6mM) or only 30 min at the high glucose concentration (83.3mM). 2. The insulinotropic action of theophylline cannot be attributed to contamination of the islets with exogenous glucose and is not suppressed by mannoheptulose. 3. The secretory response to theophylline is an immediate phenomenon, but disappears after 60min of exposure to the drug. 4. The release of insulin evoked by theophylline is abolished in calcium-depleted media containing EGTA. Theophylline enhances the net uptake of 45Ca by the islets. 5. Glycogen accumulates in the islets during the preincubation period, as judged by both ultrastructural and biochemical criteria. Theophylline significantly increases the rate of glycogenolysis during the final incubation in the glucose-free medium. 6. The theophylline-induced increase in glycogenolysis coincides with a higher rate of both lactate output and oxidation of endogenous 14C-labelled substrates. 7. These data suggest that stimulation of glycolysis from endogenous stores of glycogen is sufficient to provoke insulin release even in glucose-deprived islets, as if the binding of extracellular glucose to hypothetical plasma-membrane glucoreceptors is not an essential feature of the stimulus-secretion coupling process.