We compared three strategies for diagnosing carpal tunnel syndrome: using a single test; requiring one, two, or three of three tests to be abnormal; or utilizing a single summary variable incorporating data from three tests. Sixty-five hands of subjects without clinical carpal tunnel syndrome (CTS) were compared with 66 hands with clinical CTS. Three latency differences were measured: median-ulnar (8 cm) midpalmar orthodromic (palmdiff); median-ulnar ring finger (14 cm) antidromic (ringdiff); and median-radial thumb (10 cm) antidromic (thumbdiff). The combined sensory index (CSI) was the sum of these three differences. Sensitivity for the tests was palmdiff 69.7%, ringdiff 74.2%, thumbdiff 75.8%, and CSI 83.1%. Specificity was 95.4-96.9%. Requiring one, two, or three of three tests to be abnormal yielded sensitivities of 84.8%, 74.2%, or 56.1%, respectively, but specificities of 92.3%, 98.5%, and 100%, respectively. We conclude that a combined index improves diagnostic classification over use of single test results.