A double-blind, prospective, randomised, cross-over clinical trial was performed comparing a glibenclamide (G) 5.0 mg/metformin (M) 400 mg combination with a G 2.5 mg/M 400 mg formulation to evaluate whether a higher dose of glibenclamide was able to improve glycaemia in poorly controlled Type 2 diabetic patients. One hundred and ninety-eight patients with poorly controlled Type 2 diabetes mellitus were randomised to receive one of the two trial drugs for a first 3-month period, and were then assigned to the alternative combination for further 3 months. The starting dose (2 tablets/day, 30 min before breakfast and dinner) was to be up-titrated to 3 tablets/day when required. A standard dietary regimen was kept constant for the total trial duration. Fasting plasma glucose, HbA1c, C-peptide, insulin and lactate levels, haematology and blood chemistry were measured at the start/end of each cycle. Patients' self-assessment of the glycaemic profile (at fasting and 2 hr after the main meals) was performed weekly. Patients were constantly monitored for adverse events and episodes of hypoglycaemia, and all events were recorded. Decrease of mean fasting glucose levels measured in the first cycle was more pronounced in the group treated with G 5.0 mg/M 400 (p<0.01) compared to baseline, although the difference was not significant--no changes were observed in the second 3-month period. Results of patients' self-assessment of the glycaemic profile in the overall 6-month period show that the two trial drugs produced similar effects on fasting glucose, but the decrease of post-prandial glycaemic levels was markedly higher with G 5 mg/M 400 mg than with G 2.5 mg/M 400 mg at both main meals. A similar significant decrease (p<0.01) of HbA1c was observed in both sequence groups at the end of the first 3-month treatment period, and mean levels remained unchanged at 6 months. Drug-related adverse events were observed in 2 patients during treatment with G 2.5 mg/M 400 mg and in 5 with G 5 mg/M 400 mg, while 14 and 22 episodes of hypoglycaemia occurred with the two trial drugs, respectively (p=NS between treatments). Metformin-induced increases of lactate levels were similar in the two sequence groups. No differences between groups were found either in the number of up-titrated patients or in all the other laboratory parameters. In conclusion, the new combination containing 5-mg glibenclamide produced a greater improvement in post-prandial glycaemic control compared with the standard fixed doses, and resulted equally safe and well tolerated.