Patients with type 2 diabetes often have impaired exercise capacity compared with nondiabetic subjects. Left ventricular (LV) diastolic dysfunction has been shown to limit exercise performance in nondiabetic subjects. Men with well-controlled type 2 diabetes were divided into 2 groups: normal LV diastolic function (group 1, n = 9) or LV diastolic dysfunction (group 2, n = 10) based on standard echocardiographic criteria using pulmonary veins and transmitral flow recordings. They were matched for age and had no evidence of systemic hypertension, macroalbuminuria, coronary artery disease, congestive heart failure, clinical diabetic complications, and thyroid disease. Good metabolic control was demonstrated by glycated hemoglobin levels of 6.7+/-1.6% and 6.6+/-2.5% (means +/- SD) in patients with LV diastolic dysfunction and in controls, respectively. Each subject performed a symptom-limited modified Bruce protocol treadmill exercise test. Maximal treadmill performance was higher in subjects with normal diastolic function compared with subjects with LV diastolic dysfunction when expressed in time (803+/-29 vs. 662+/-44 seconds, respectively, p<0.02) or in METs (11.4+/-1.2 vs. 9.5+/-1.9 METs, respectively, p<0.02). Moreover, there was a correlation between E/A ratio and exercise duration (r = 0.64, p = 0.004) or E/A ratio and METs (r = 0.658, p = 0.003). There were no significant differences in maximal heart rate, maximal systolic and diastolic blood pressure, or maximal rate-pressure product attained during the exercise test. In conclusion, this study demonstrated that LV diastolic dysfunction influences maximal treadmill performance and could explain lower maximal performance observed in patients with type 2 diabetes.