Abstract Aim Incipient cardiovascular autonomic imbalance is not readily diagnosed by conventional methods. Spectral analysis of heart rate variability (HRV) by wavelet transform (WT) was used to measure cardiovascular autonomic function in patients with Type 2 diabetes. Methods Thirty-two diabetic patients without (D), 26 with cardiovascular autonomic neuropathy (DAN) and 72 control subjects (C) participated. A 30-min HRV time series was analysed by wavelet transformation and four characteristic frequency intervals were defined: I (0.0095–0.021 Hz), II (0.021–0.052 Hz), III (0.052–0.145 Hz) and IV (0.145–0.6 Hz). Results When compared with C, in both D and DAN the normalized power and amplitude of interval II were increased and of interval IV decreased, resulting in a significantly higher II/IV ratio. Furthermore, in DAN the normalized power and amplitude of interval I were increased and of interval III decreased when compared with the D and C groups. The diabetic patients were divided in two equal subgroups according to HbA1c < 8.0% and ≥ 8.0%. In the subgroup with HbA 1c ≥ 8.0%, normalized power in interval II was significantly higher and in interval IV significantly lower than in the subgroup with HbA 1c < 8.0%. In D, but not in DAN patients prescribed ACE inhibitors, the absolute amplitude and power of oscillations were significantly higher than in patients not taking ACE inhibitor therapy. Conclusions Patients with diabetes have increased sympathetic and decreased parasympathetic cardiac activity regardless of the presence of autonomic neuropathy. Glycaemic control and treatment with ACE inhibitors may favourably influence HRV in diabetic patients without autonomic neuropathy.