Heart rate variability (HRV) is recognized to carry early diagnostic value regarding cardiac autonomic neuropathy (CAN). A number of different HRV analysis algorithms have been proposed for the assessment of CAN, each of them providing partly differing information about HRV time series. Instead of confining to a limited set of HRV features, a multi-dimensional approach incorporating a multitude of HRV parameters could be an optimal way of assessing the changes in HRV related to CAN. In this paper, principal component analysis (PCA) is used for analysing multi-dimensional HRV data of 11 patients with definite CAN and 71 subjects without CAN. Using the two most significant principal components, patients with CAN were separated from subjects without CAN with 87% accuracy.