The effects of thiazolidinediones on cardiac function are controversial in humans with type 2 diabetes (T2DM) and in animals. Given the high prevalence and prognostic relevance of diastolic myocardial dysfunction in T2DM, we tested the hypothesis that by reducing oxidative stress rosiglitazone, but not glimepiride, may improve diastolic function. This randomised cross-over study investigated 12 metformin-treated T2DM patients without cardiovascular disease before and after 16 weeks of additional therapy with rosiglitazone (8 mg daily) or glimepiride (3 mg daily). Systolic and diastolic myocardial velocity (E') were assessed with tissue Doppler. In spite of similar non-significant lowering of glycosylated haemoglobin (HbA1C), rosiglitazone, but not glimepiride, significantly improved E' (p=0.04), reduced malondialdehyde (p=0.028), lowered high-sensitivity C-reactive protein (hsCRP) (p=0.019), and increased adiponectin (p=0.002). For rosiglitazone, multivariate regression analysis revealed malondialdehyde reduction as an independent determinant of treatment-induced improvement in E'. The rosiglitazone-induced improvements of diastolic function and oxidative stress may be of prognostic relevance in choosing therapy for T2DM patients without overt heart disease.