OBJECTIVES: Pioglitazone (PIO), a thiazolidinedione, is a member of a new class of oral antidiabetic agents targeted to treat insulin resistance, the major underlying cause of type 2 diabetes mellitus. Insulin (INS) is believed to be the "gold standard" for achieving optimal glycaemic control. This study is aimed to compare both treatment options with regard to metabolic control and effectiveness under the conditions of daily practice. METHODS: Prospective, controlled, non-randomized observational study where patient selection, allocation to treatment and dose was left at the physicians' discretion. Quality standards included a central laboratory and a regular monitoring. Primary parameter was the change of HbA1C compared to baseline (? HbA1C) where a difference of <0.5% between both arms was set for defining non-inferiority. Analyses were performed under the perspective of the German Statutory Health Care System. RESULTS: A total of 299 and 218 patients in whom PIO or INS was started for insufficient metabolic control were treated for a mean duration of 26 weeks at 51 specialised out-patient diabetic centres. Adjusted ? HbA1C (- 0.72), PIO; (- 0.42) INS) and adjusted ? Fasting Plasma Glucose - 24.5 mg/dl, PIO; - 19.5 mg/dl, INS) were observed. Responder rates (adjusted ? HbA1C = 0.6%) were 54.9% (PIO) and 37.2% (INS), respectively. Mean total treatment costs were €1207 (PIO) and €1510 (INS) where mean costs for antidiabetic medication and glucose self-monitoring (dip-stick measurement) could be assessed as €646 (PIO) and €774 (INS). Compared to INS, PIO revealed to be more cost-effective (? HbA1C/€ 1000) in the insulin resistant (ATP-III), younger (<64 years) and more obese (BMI > 30 kg/m2) individuals with shorter diabetes duration (<5 years). For INS a similar trend could not be observed. CONCLUSIONS: PIO proved to be non-inferior to INS treatment in terms of metabolic control as well as cost-effectiveness. Targeting individual patient profiles achieve best possible outcomes.