The validity of twelve provocative tests for carpal tunnel syndrome (CTS) in a random sample of 504 people from the general population was assessed. 50 woke up at night due to paraesthesiae (with or without numbness or pain) in the fingers innervated by the median nerve (CTS symptoms) in 93 hands. CTS was neurophysiologically confirmed in 28 subjects (44 hands)--a prior probability for CTS of 47%. All clinical diagnostic tests had a low validity. Posterior probability of CTS ranged from 35 to 70% for positive test results and from 41 to 62% for negative test results. A combination of three tests with relatively high validity (paresis of abductor pollicis brevis muscle, hyperpathia, and flick sign) did not significantly change the probability of CTS. Patients with CTS symptoms should be referred directly for neurophysiological examination.