Aims This study was designed to test the efficacy and safety of low-dose rosiglitazone, a potent, insulin-sensitizing thiazolidinedione, in combination with sulphonylurea in Type 2 diabetic patients. Methods For the intention-to-treat analysis, 574 patients (59% male, mean age 61 years) were available, randomized to receive 26 weeks of twice-daily placebo (n = 192), rosiglitazone 1 mg (n = 199) or rosiglitazone 2 mg (n = 183) in addition to existing sulphonylurea treatment with gliclazide (47.6% of patients), glibenclamide (41.8%) or glipizide (9.4%) (two patients were taking carbutamide or glimepiride). Change in haemoglobin A(1c) (HbA(1c)), fasting plasma glucose (FPG), frustosamine, insulin, C-peptide, albumin, and lipids were measured, and safety was evaluated. Results Mean baseline HbA(1c) was 9.2% and FPG was 11.4 mmol/l. Rosiglitazone at doses of 1 and 2 mg b.d, plus sulphonylurea produced significant decreases, compared with sulphonylurea plus placebo, in HbA(1c) (-0.59% and -1.03%, respectively; both P <0.0001) and FPG (1.35 mmol/l and 2.44 mmol/l, respectively; both P <0.0001). Both HDL-cholesterol and LDL-cholesterol increased and potentially beneficial decreases in non-esterified fatty acids and gamma glutamyl transpeptidase levels were seen in both rosiglitazone groups. The overall incidence of adverse experienccs was similar in all three treatment groups, with no significant cardiac events, hypoglycaemia or hepatotoxicity. Conclusions Overall, the combination of rosiglitazone and a sulphonylurea was safe, well tolerated and effective in patients with Type 2 diabetes.