The changes in the regulation of insulin secretion that accompany sepsis are yet to be fully established. We therefore examined insulin secretion both in vivo and in vitro in 2 different models of peritonitis/sepsis in the rat. Sepsis was induced by intraperitoneal injection of Escherichia coli either alone or together with bile. Following sepsis induction, an initial hyperglycemia developed. This hyperglycemia was transient and had vanished after 3 h (coli group) or 9 h (bile group). However, after 24 h, a second phase of hyperglycemia developed in both groups. The glucose elimination rate after intravenous glucose injection (0.5 g/kg) at 4 and 10 h after peritonitis/sepsis induction was retarded and the hyperglycemia that occurred during intravenous glucose infusion (10 mg/min for 30 min) was exaggerated. This is consistent with a reduced glucose uptake. Simultaneously, the plasma insulin responses to glucose were markedly exaggerated. This could be due to a true potentiated insulin secretion or simply to an adaptation to the hyperglycemia. However, also during intravenous arginine infusion (7 mg/min) at 4 h after peritonitis/sepsis induction, the plasma insulin responses were markedly exaggerated. Since only a slight change in plasma glucose occurred during this challenge, the results suggest that sepsis is accompanied by a true hypersecretion of insulin. To verify whether this is directly or indirectly mediated, pancreatic islets were isolated from peritonitis/sepsis animals at 4 h after disease induction and incubated for 45 min in a KRB medium supplemented with different concentrations of glucose. The subsequent insulin secretion was the same in islets from the septic animals as in controls. Hence, our results show that experimental peritonitis/sepsis in the rat is accompanied by (1) glucose intolerance and (2) a true hypersecretion of insulin which is indirectly mediated.