Which therapy should be used in Type 2 (non-insulin-dependent) diabetic patients with "secondary sulfonylurea failure", insulin or a combination of sulfonylurea and metformin? To address this question, we have compared the effect of 6 months of insulin therapy twice daily with that of a combination of glibenclamide and metformin in 24 Type 2 diabetic subjects, who no longer responded to treatment with sulfonylureas. Both treatments resulted in an equivalent 30% improvement in mean daily blood glucose (p less than 0.001), without significant effect on serum lipids. Insulin improved glycaemic control primarily by reducing basal hepatic glucose production (p less than 0.05), but had no significant effect on peripheral glucose metabolism. The combination of glibenclamide and metformin enhanced significantly total body glucose metabolism (p less than 0.05), predominantly by stimulating the non-oxidative pathway. Neither insulin nor the combination therapy altered B-cell response to a test meal. Insulin therapy resulted in a 6% increase in body weight, 63% of which was accounted for by increased fat mass. Although body weight was unchanged during sulfonylurea/metformin therapy, lean body mass and energy expenditure decreased significantly (p less than 0.05). We conclude that insulin and glibenclamide/metformin have different long-term effects on glucose and energy metabolism in Type 2 diabetes.