Background: Diagnosis of carpal tunnel syndrome (CTS) is based on clinical symptoms, examination findings, and electrodiagnostic studies (EDX). For CTS, the most useful part is nerve conduction studies (NCSs). However, NCS can result in ambiguous or false negative results, particularly for mild CTS. Increasing the number of NCS tests improves accuracy, but also increases time, cost, and discomfort. To improve accuracy without additional testing, the Terminal Latency index (TLI) and Residual Latency (RL) are additional calculations using the minimum number of tests. Recently, the median sensory-ulnar motor latency difference (MSUMLD) was devised as another way to improve diagnostic accuracy for mild CTS.Aims: MSUMLD, TLI and RL were compared for diagnostic accuracy by severity.Study Design: Diagnostic accuracy studyMethods: A total of 657 subjects were retrospectively enrolled. The CTS group consisted of 546 subjects with CTS by NCS (all severities). The control group consisted of 121 subjects with no hand symptoms and normal NCS. All statistical analyses were performed using SAS. Means were compared using the one-way ANOVA test with the Bonferroni adjustment. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were compared including Receiver Operating Characteristic (ROC) curve analysis.Results: For mild CTS, MSUMLD showed the higher specificity and PPV rates (0.967 and 0.957, respectively) than TLI (0.603 and 0.769, respectively) and RL (0.818 and 0.858, respectively). The area under the ROC was highest for the MSUMLD (0.889), followed by the RL (0.829), and the TLI last (0.762). Differences were statistically significant (MSUMLD is the most accurate). For moderate CTS, sensitivity and specificity rates of RL (0.989 and 1.000) and TLI (0.983 and 0.975) were higher than MSUMLD (0.866 and 0.958). Differences in area under the ROC curve were not significantly significant, but MSUMLD sensitivity was lower. For severe CTS, RL yielded 1.000 sensitivity, specificity, PPV, NPV and area beneath ROC Curve. Differences in area under the ROC curve were not significantly significant.Conclusion: MSUMLD is the best calculated parameter for diagnosing mild CTS. It requires just a simple calculation and no additional testing. RL and TLI are also useful in diagnosing mild to moderate CTS.