OBJECTIVE Cardiac autonomic neuropathy (CAN) as a result of diabetic autonomic neuropathy is positively related to a poor prognosis in diabetic patients. The measurement of heart rate variability (HRV) is a remarkable index of cardiac autonomic dysfunction. The aim of this study was to examine the effects of long-term exercise training on HRV in type 2 diabetic patients with definite CAN. METHODS Seventeen type 2 diabetic patients with definite CAN (group A: 56.2 years (SD 5.8)) and 15 without CAN (group B: 55.8 years (SD 5.6)) participated in the study. All patients followed an aerobic exercise training programme three times a week for 6 months; the intensity of the session was 70% to 85% of heart rate reserve. At the beginning and end of the study all subjects underwent graded maximal exercise testing with spiroergometry for the evaluation of their aerobic capacity (VO(2)peak). Moreover, time and frequency domain indices of HRV were obtained from 24 h ambulatory continuous ECG Holter recordings. RESULTS At baseline, all measurements of HRV indices were significantly reduced in group A compared with group B (p<0.05). Moreover, group A reached a significantly lower VO(2)peak by 14.8% compared with group B (p<0.05). Following the exercise training programme, the SD of all normal-to-normal RR intervals in the entire recording (SDNN) was increased by 18.8% (p<0.05) and 13.8% (p<0.05), the square root of the average of sum of squares of difference between adjacent filtered RR intervals (rMSSd) was increased by 35% (p<0.05) and 15.2% (p<0.05), and the percentage of differences between adjacent filtered RR intervals which was greater than 50 ms for the entire analysis (pNN50) was increased by 400% (p<0.05) and 67.9% (p<0.05) in groups A and B, respectively. Regarding the frequency domain indices, only the high frequency power (HF) was found to be significantly increased in group A. At the end of the exercise training programme, SDNN, rMSSd and low frequency power (LF) were significantly lower (24.3% (p<0.05), 20.3% (p<0.05) and 40% (p<0.05), respectively) in group A compared with group B. Also, VO(2)peak increased by 17.8% (p<0.05) in group A and by 11% (p<0.05) in group B. Furthermore, the exercise training programme had significant effects on blood lipid and glucose levels and glycosylated haemoglobin (HbA(1c)) in both groups. CONCLUSIONS The results indicate that 6-month aerobic exercise training improves the cardiac autonomic nervous system function in type 2 diabetic patients. However, more favourable effects are found in type 2 diabetic patients with definite CAN.