The study of diabetic polyneuropathy is complicated by a lack of clear definitions and the absence of a simple reliable test procedure. Recently, a new sensory perception testing device has been introduced for detection of thresholds for electrical stimuli (current perception: CPT) at different frequencies (Neurometer). We compared standardized clinical examination scores with measurements of vibratory perception threshold (VPT) and CPT (foot) and obtained reproducibility figures. Participants in the study were healthy controls (H, n = 33), diabetic patients without clinical signs of neuropathy (DN-, n = 23), diabetics with overt diabetic neuropathy (DN+, n = 22), and patients with a diabetes duration of over 20 years (D20, n = 38). As expected, there were highly significant differences (Wilcoxon) in CPT, VPT and neurological scores between H/DN- and DN+ (p < 0.001), but not between H and DN-. Correlation between CPT and total as well as partial (reflecting small and large fibre functions) neurological examination score were highest at 2000 Hz (r = 0.88); no advantage of lower frequency CPT could be identified. CPT seemed rather insensitive in detecting neuropathy. Correlations between CPT and VPT were only moderate and maximal at 2000 Hz (r = 0.61). Reproducibility of CPT was good at 2000 Hz (coefficient of variation 13.3-20.2%), but moderate to poor at lower frequencies (ranging to 62%). We conclude that CPT and VPT quantitative sensory testing is only of limited value, mainly because of high variability and poor reproducibility.