The risk related to cardiovascular autonomic neuropathy dysautonomia should lead to a specific assessment of this complication of diabetes. The aim of this study was to estimate the accuracy of a battery of blood pressure (BP) and heart rate (HR) variability indexes obtained in different subgroups of diabetic subjects classified according to the conventional laboratory autonomic function tests (Ewing scores). Blood pressure was measured continuously at the finger level with a Finapres monitor while subjects were in the supine position and again while they were standing. Pulse intervals were derived from BP recordings and were taken as surrogates for R-R intervals. Subjects with borderline or definite cardiovascular autonomic neuropathy showed a similar degree of alterations of both HR and BP variability (spectral measures) and in the relationship between BP and HR (cross-spectral and sequence analysis). Subjects with no evidence of cardiovascular autonomic neuropathy on the basis of the conventional tests showed an altered relationship between BP and HR. This baroreceptor-HR reflex dysfunction could represent an early stage of cardiovascular autonomic neuropathy undetected by the conventional tests. The areas under the receiver operating characteristic plots indicated that the high-frequency peak of pulse interval was highly discriminant in the supine and standing positions. The cross-spectral analysis showed the best discrimination for the gain in the high-frequency range. For the sequence analysis, the slope was the best discriminant factor for any degree of cardiovascular autonomic neuropathy. In conclusion, these estimates of baroreceptor-HR function may provide a powerful tool for assessing cardiovascular autonomic neuropathy at any stage, including the early stage, which is not detected by the conventional tests.