Abstract Aims: Our objective was to investigate whether improved in vivo mitochondrial function in skeletal muscle and intramyocellular lipids (IMCLs) contribute to the insulin-sensitizing effect of rosiglitazone. Methods: Eight overweight type 2 diabetic patients (body mass index = 29.3 ± 1.1 kg/m2) were treated with rosiglitazone for 8 wk. Before and after treatment, insulin sensitivity was determined by a hyperinsulinemic euglycemic clamp. Muscular mitochondrial function (half-time of phosphocreatine recovery after exercise) and IMCL content were measured by magnetic resonance spectroscopy. Results: Insulin sensitivity improved after rosiglitazone (glucose infusion rate: 19.9 ± 2.8 to 24.8 ± 2.1 μmol/kg·min; P < 0.05). In vivo mitochondrial function (phosphocreatine recovery half-time: 23.8 ± 3.5 to 20.0 ± 1.7 sec; P = 0.23) and IMCL content (0.93 ± 0.18% to 1.37 ± 0.40%; P = 0.34) did not change. Interestingly, the changes in PCr half-time correlated/tended to correlate with changes in fasting insulin (R2 = 0.50; P = 0.05) and glucose (R2 = 0.43; P = 0.08) levels. Changes in PCr half-time did not correlate with changes in glucose infusion rate (R2 = 0.08; P = 0.49). Conclusion: The rosiglitazone-enhanced insulin sensitivity does not require improved muscular mitochondrial function.