We determined the inter- and intraexaminer reliability of nerve conduction measurements in six patients with diabetic peripheral neuropathy. Each patient was examined by six electromyographers on two separate occasions at least 1 week apart. We obtained attributes of nerve conduction at each examination and analyzed the data by analysis of variance. Intraexaminer reliability was high for 11 of 12 measurements, and interexaminer reliability was high for eight of twelve. Three of the four measurements that varied between examiners were either sensory or motor amplitudes, attributes frequently used to measure disease progression or to assess the result of therapeutic intervention. Our results suggest that longitudinal nerve conduction measurements used to assess worsening or improvement over time should optimally be performed by a single examiner to minimize the degree of variability associated with different examiners.