Background. Several efficacy studies of insulin-therapy regimens in patients with type 2 diabetes mellitus have shown varying results. Moreover, most studies did not address hypoglycaemia frequency and severity. Methods. In this multicentre study, we compared the glycaemic efficacy and incidence rate of hypoglycaemic episodes between 3 treatment regimens in obese type 2 diabetic patients with secondary failure to sulphonylurea and metformin. During the run-in phase, patients were treated with glimepiride and metformin. After 3 months, 261 patients with HbA1c values >6.5% were randomised to (A) glimepiride + Neutral Protein Hagedorn (NPH) insulin at bedtime, (B) NPH insulin twice daily and (C) 30/70 mixture of short-acting and NPH insulin twice daily. The therapeutic aim was an HbA1c level ≤6.5%. Results. Mean HbA1c achieved at 9 months was significantly higher in group A: 8.9% versus 8.3% and 8.4% in groups B and C, respectively (P < 0.001). There was no difference in the mild hypoglycaemic event rate, 0.36 versus 0.48 versus 0.53 events per patient month, in groups A, B and C, respectively. Severe hypoglycaemic events, requiring help from others, did not occur throughout the study. The mean weight gain and insulin dose were comparable in all three groups. Conclusions. The glimepiride + NPH insulin treatment resulted in a higher HbA1c level, as compared to the other regimens. In the clinical setting of this multicentre study, good glycaemic control was only achieved in a minority of the patients, irrespective of the applied regimen.