Thirteen patients (6 females and 7 males) who were secondary failures on oral drug therapy were randomly allocated to either 2 months of treatment with insulin + glibenclamide or insulin + placebo. Thereafter the treatment schedules of the two groups were switched over for another two months. The combination of insulin and glibenclamide was more effective in lowering the fasting blood glucose (P = 0.026) and 24 h urine glucose (P = 0.042) than the combination of insulin and placebo. The combination therapy with insulin and glibenclamide revealed higher basal (P = 0.021) and glucagon-stimulated C-peptide concentrations (P = 0.037) than therapy with insulin and placebo. However, insulin binding to erythrocytes did not differ between the two study periods. The results indicate that the addition of glibenclamide to insulin in type II diabetics poorly controlled by oral antidiabetics alone may slightly improve diabetic control. The mechanism of this action is due at least partly to sulphonylurea-induced stimulation of endogenous insulin secretion. The effectiveness of the combination treatment during long-term therapy still remains to be proven, however.