1. To determine how pretreatment with sulphonylureas alters the beta cell function, mouse islets were cultured (18 - 20 h) without (controls) or with (test) 0.01 microM glibenclamide. Acute responses to glucose were then determined in the absence of glibenclamide. 2. Test islets were insensitive to drugs (sulphonylureas and diazoxide) acting on K+-ATP channels, and their [Ca2+]i was already elevated in the absence of stimulation. 3. Insulin secretion was increased in the absence of glucose, and mainly stimulated between 0 - 10 instead of 7 - 20 mM glucose in controls. The maximum response was halved, but this difference disappeared after correction for the 45% decrease in the islet insulin content. 4. The first phase of glucose-induced insulin secretion was abrogated because of a paradoxical decrease of the high basal [Ca2+]i in beta cells. The second phase was preserved but occurred with little rise of [Ca2+]i. These abnormalities did not result from alterations of glucose metabolism (NADPH fluorescence). 5. In islets cultured with 50 microM tolbutamide, glucose induced biphasic increases in [Ca2+]i and insulin secretion. The decrease in the secretory response was matched by the decrease in insulin content (45%) except at maximal glucose concentrations. Islets pretreated with tolbutamide, however, behaved like those cultured with glibenclamide if tolbutamide was also present during the acute functional tests. 6. In conclusion, treatment with a low glibenclamide concentration causes long-lasting blockade of K+-ATP channels and rise of [Ca2+]i in beta cells. Glucose-induced insulin secretion occurs at lower concentrations, is delayed and is largely mediated by a modulation of Ca2+ action on exocytosis. It is suggested that glucose regulation of insulin secretion mainly depends on a K+-ATP channel-independent pathway during in vivo sulphonylurea treatment.