Carpal tunnel syndrome (CTS) is the most common neurologic entrapment disorder diagnosed in the upper limb. Nevertheless, there is still debate about the most reliable test that should be performed to diagnose CTS. Much of the argument has been drawn from the opinions of individuals or groups with varying degrees of expertise in the field; little has been based on actual data. The purpose of this study was to investigate the diagnostic patterns of CTS in an academic medical setting in southern Taiwan. The charts of 1,050 patients with a diagnosis of CTS over a 1-year period (2001-2002) were retrospectively reviewed. Data on 622 patients with new-onset CTS were included in the analysis. On the patient's initial visit, physicians made a diagnosis of CTS in 34.9% of cases solely on the basis of the history of symptoms without resort to provocative tests, while 8.7% of cases were diagnosed on the basis of symptom characteristics alone in spite of negative provocative tests. A CTS diagnosis was given according to symptoms and positive provocative tests in 55% of cases. Apart from these, CTS diagnosis remained unchanged in 27.3% of cases without electrodiagnostic signs of CTS during follow-up visits. An average of 1.6 diagnostic maneuvers were conducted for CTS patients, with nerve conduction velocity (NCV) studies (516 cases) being the most frequently performed, followed by Tinel's sign (350 cases) and Phalen's test (102 cases). Our findings imply that physicians are inclined to base their diagnosis on clinical history and physical examination for patients with suspected CTS. Clear guidelines regarding the indications for referral for NCV studies should be established in response to the increased concerns about the cost effectiveness of diagnostic tests.