The aim of the study was to evaluate the long-term effect of rosiglitazone and metformin monotherapy with medical nutrition treatment (MNT) and of MNT alone on arterial stiffness, serum monocyte chemoattractant protein (MCP)-1 and matrix metalloproteinase (MMP)-9 in drug naive patients with type 2 diabetes mellitus. Fifty type 2 diabetic patients were randomized to receive rosiglitazone 4 mg/day (n=19) or metformin 850 mg/day (n=16) with MNT or MNT alone (n=15), for 52 weeks. Arterial stiffness was assessed by using large and small artery elasticity index (SAEI and LAEI, respectively). SAEI, LAEI, serum MCP-1 and MMP-9 levels were measured at baseline and following 52 weeks of treatment. SAEI was improved only in the rosiglitazone group, and the difference was still statistically significant when the three groups were compared (p=0.024). There were no differences in LAEI in inter- and intragroup comparisons at the end of the study. Serum MMP-9 levels were decreased in the metformin (-13.5+/-34.8%, p=0.02) and rosiglitazone (-27.2+/-51.0%, p=0.023) groups compared with baseline values, whereas no significant change was seen in serum MCP-1 levels. These results suggest that rosiglitazone monotherapy has favorable effects on arterial stiffness compared with metformin monotherapy independent of glycemic control.