OBJECTIVE: Autonomic dysfunction may contribute to the etiology and exercise intolerance of subclinical diabetic heart disease. This study sought the efficacy of exercise training for improvement of peak oxygen uptake (VO₂(peak)) and cardiac autonomic function in type 2 diabetic patients with non-ischemic subclinical left-ventricular (LV) dysfunction. MATERIALS/METHODS: Forty-nine type 2 diabetic patients with early diastolic tissue Doppler velocity >1 standard deviation below the age-based mean entered an exercise intervention (n=24) or usual care (n=25) for 6-months (controlled, pre-/post- design). Co-primary endpoints were treadmill VO₂(peak) and 5-min heart-rate variability (by the coefficient of variation of normal RR intervals [CVNN]). Autonomic function was additionally assessed by resting heart-rate (for sympathovagal balance estimation), baroreflex sensitivity, cardiac reflexes, and exercise/recovery heart-rate profiles. Echocardiography was performed for LV function (systolic/diastolic tissue velocities, myocardial deformation) and myocardial fibrosis (calibrated integrated backscatter). RESULTS: VO₂(peak) increased by 11% during the exercise intervention (p=0.001 vs. -1% in controls), but CVNN did not change (p=0.23). Reduction of resting heart-rate in the intervention group (p<0.05) was associated with an improvement in the secondary endpoint of heart-rate variability total spectral power (p<0.05). However, baroreflex sensitivity, cardiac reflexes, and exercise/recovery heart-rate profiles showed no significant benefit. No effects on LV function were observed despite favorable reduction of calibrated integrated backscatter in the intervention group (p<0.05). CONCLUSIONS: The exercise intolerance of subclinical diabetic heart disease was amenable to improvement by exercise training. Despite a reduction in resting heart-rate and potential attenuation of myocardial fibrosis, no other cardiac autonomic or LV functional adaptations were detected.