Increasingly more tests are being used to detect and characterize diabetic polyneuropathy, but their value in setting minimal criteria for the diagnosis of neuropathy and for staging severity remains inadequately studied. In 180 diabetics, we compared the percentage of patients with test abnormalities and associations among test results, evaluating neuropathic symptoms [neuropathy symptom score (NSS) and neuropathy scale of neuropathy symptom profile (NNSP)], deficits [neurologic disability score (NDS) and vibratory (VDT) and cooling (CDT) detection thresholds], or nerve dysfunction [nerve conduction (NC)]. The percentage of patients that were abnormal varied considerably depending on criteria for abnormality and the tests used. Abnormality (greater than or equal to 3 SD of 1 or more parameters) of NC of one or more of four nerves occurred in 80%, of two or more in 69%, of three or more in 46%, and of four in 21%. Similarly, for other tests, the rate of abnormality decreased with use of increasingly stringent criteria. Setting the criteria for abnormal NC at abnormality of two or more nerves, NSS at greater than or equal to 1, NDS at greater than 6, NNSP at greater than or equal to 97.5th percentile, and at greater than or equal to 95th percentile for the other tests, NC was abnormal in 69%, NSS in 54%, NDS in 48%, NNSP in 47%, VDT in 44%, and CDT in 35%. Abnormality of any two or more of the six tests evaluated occurred in 64% of patients.(ABSTRACT TRUNCATED AT 250 WORDS)