This 6-month, double-blind, placebo-controlled, randomised, parallel-group study investigated the potential of acarbose add-on therapy for improving the glycaemic control of overweight patients with Type 2 diabetes and was inadequately controlled with metformin monotherapy. Patients were randomised to receive acarbose titrated up to 100 mg three times daily (n=74) or placebo (n=78). All patients were receiving metformin 850 mg twice or thrice daily before the study and continued to receive this dose throughout the study. The mean difference in glycated haemoglobin (HbA(1c)) (+/-S.D.) from baseline to endpoint was -0.7+/-1.2% U in the acarbose intention-to-treat (ITT) group, compared with +0.2+/-1.3% in the placebo ITT group (P=0.0001). Significantly, more patients in the acarbose group were classified as 'responders', with an HbA(1c) at the end of treatment of less than 7.0% or a decrease by at least 15% relative to baseline (acarbose vs. placebo; 42 vs. 17%; P=0.002). The difference in fasting blood glucose level from baseline to endpoint was -1.0+/-2.8 (S.D.) mmol/l in the acarbose ITT group, compared with +1.3+/-2.8 mmol/l in the placebo ITT group (P=0.0001), and for 2-h postprandial blood glucose level -1.4+/-3.8 vs. +1.1+/-3.5 mmol/l (P=0.0001). In all, 60% of patients in the acarbose group and 33% in the placebo group had an adverse event considered to be possibly or probably related to drug therapy, leading to withdrawal by 15 and 3%, respectively. The results indicate that acarbose has potential clinical utility for improving glycaemic control in overweight patients with Type 2 diabetes inadequately controlled with metformin.