Glucose-induced insulin secretion is enhanced by a preceeding glucose stimulus. The characteristics of this action of glucose were investigated in perfused pancreas and collagenase-isolated islets of Langerhans. A 20- to 30-min pulse of 27.7 mM glucose enhanced both the first and second phase of insulin release in response to a second glucose stimulus by 76-201%. This enhancement was apparent as an augmented maximal insulin release response to glucose. The effect of priming with glucose was seen irrespective of whether the pancreatic tissue was obtained from fed or fasted rats. Separating the two pulses of hexose by a 60-min time interval of exposure to 3.3 mM glucose did not abolish the potentiation of the second pulse. Omission of Ca(++) as well as the inclusion of somatostatin or mannoheptulose during the first pulse abolished insulin secretion during this time period; however, only the inclusion of mannoheptulose deleted the potentiation of the second pulse. d-Glyceraldehyde, but not pyruvate, d-galactose, or 3-isobutyl-1-methylxanthine, could substitute for glucose in inducing potentiation. In islets labeled with [2-(3)H]adenine, the [(3)H]cyclic AMP response to glucose was increased by 35% when measured after 1 min, but was increased only marginally after 2-10 min of stimulation with a second pulse of glucose. The production of (3)H(2)O from glucose was not affected by glucose priming. It is concluded that (a) the induction of the glucose-induced, time-dependent potentiation described here is dependent on glucose metabolism but not on stimulation of cyclic AMP, calcium fluxes, or insulin release per se; (b) the mechanisms that mediate the pancreatic "memory" for glucose are unknown but do not seem to involve to a major extent an increased activity of the adenylate cyclase-cyclic AMP system of the beta-cell; (c) the evidence presented supports the hypothesis of a dual role of glucose for insulin release.