OBJECTIVE: To evaluate whether treatment with insulin in recently diagnosed type 2 diabetes is advantageous compared with glibenclamide treatment. RESEARCH DESIGN AND METHODS: Beta-cell function, glycemic control, and quality of life were monitored over 2 years in 39 patients with islet cell antibody-negative type 2 diabetes diagnosed 0-2 years before inclusion in a Swedish multicenter randomized clinical trial. Patients were randomized to either two daily injections of premixed 30% soluble and 70% NPH insulin or glibenclamide (3.5-10.5 mg daily). C-peptide-glucagon tests were performed yearly in duplicate after 2-3 days of temporary withdrawal of treatment. RESULTS: After 1 year the glucagon-stimulated C-peptide response was increased in the insulin-treated group by 0.14 +/- 0.08 nmol/l, whereas it was decreased by 0.12 +/- 0.08 nmol/l in the glibenclamide group, P < 0.02 for difference between groups. After 2 years, fasting insulin levels were higher after treatment withdrawal in the insulin-treated versus the glibenclamide-treated group (P = 0.02). HbA(1c) levels decreased significantly during the first year in both groups; however, at the end of the second year, HbA(1c) had deteriorated in the glibenclamide group (P < 0.01), but not in the insulin-treated group. The difference in evolution of HbA(1c) during the second year was significant between groups, P < 0.02. A questionnaire indicated no difference in well-being related to treatment. CONCLUSIONS: Early insulin versus glibenclamide treatment in type 2 diabetes temporarily prolongs endogenous insulin secretion and promotes better metabolic control.