Nerve conduction studies performed between 1994 and 1996 on the median nerves of 204 hands of 113 patients [20 patients in hemodialysis (HD) with diabetic nephropathy, 27 without diabetes mellitus in HD for 10 years or more, 18 without diabetes mellitus in HD for less than 10 years, and 48 with diabetes mellitus not undergoing HD] were reviewed retrospectively to develop a discriminating tool that would better inform treatment selection for patients with carpal tunnel syndrome (CTS) in HD. Scatter diagrams were drawn from the results of this study, and a regression formula was obtained. Among the long-term HD patients without diabetic nephropathy, distal motor latency (DML) tended to increase without a significant concomitant reduction in motor nerve conduction velocity (MCV) in the forearm. In contrast, in HD patients with diabetic nephropathy, DML tended to increase and MCV was significantly lower despite shorter HD periods. The study showed that in HD patients, in whom the MCV(m/s) > 11[9 - DML(ms)], the surgical results might be excellent or good. The formula was tested against the results of the 30 surgical cases. The sensitivity of this formula was 0.84, and the specificity was 1.00. The formula produced a predictor of surgical results when used in conjunction with other criteria for CTS in HD patients, especially those with diabetic nephropathy.