In order to analyse further the pathophysiology of pentamidine effects on blood glucose regulation, the following experimental models were established in rats: impairment of the renal function, bile duct ligation, inhibition of the P450 cytochrome enzyme system. In otherwise intact rats, 7.5 mg/day pentamidine was well tolerated whereas doses of 15 mg/day induced severe, relapsing and eventually lethal hypoglycaemia within a few days. Induction of a renal insufficiency of graded severity by treatment with gentamycin, subtotal nephrectomy and total bilateral nephrectomy resulted in repetitive, severe (sometimes lethal) hypoglycaemia, alternating with hyperglycaemia, glucosuria and ketonuria in pentamidine-treated rats (7.5 mg/d). No long-standing insulin-dependent diabetes was observed. In the dysglycemic animals, plasma insulin levels were inappropriate to the concomitant glycaemia; no stimulation was obtained by i.v. glucose. Glucagon levels were higher than normal, suppressible by i.v. glucose, responsive to IV arginine and to hypoglycaemia. Dysglycemic events were more frequent and marked in the rats with the most severe renal functional derangement. They were more frequent in the rats treated with pentamidine mesylate than in those treated with the isethionate salt. Control uremic rats (free of pentamidine) remained euglycaemic. The islets of Langerhans displayed severe vascular congestion and degranulation and necrosis of the B cells, while the non B cells (and particularly the A cells) were intact. Exocrine pancreatitis was occasionally observed in the most severely uremic rats. In contrast with uremic rats, neither surgical ligation of choledocus, nor treatment by P450 cytochrome inhibitors (particularly ketoconazole) precipitated dysglycaemia in the pentamidine-treated rats. These experimental data: 1) strengthen the concept of inappropriate insulin release from pentamidine-lesioned islet B cells due to pentamidine accumulation; 2) indicate a predominant role for renal insufficiency in determining the accumulation of this drug; 3) emphasize the clinical importance of renal insufficiency as a risk factor for pentamidine-induced dysglycaemia. Association with ketoconazole does not appear to be a risk factor.