Clinical tests of hearing are regularly used in adults but their role, now that pure-tone audiometry is almost universally available, has not been evaluated by modern methods of analysis including sensitivity and specificity. Free-field voice testing was carried out in 101 patients and the Rinne tuning-fork test in a different group of 127 patients prior to clinical or audiometric evaluation. The results were subsequently compared to air and bone conduction pure-tone thresholds assessed using rigorous standards. Depending on the audiometric definition as to what constitutes a hearing impairment, the sensitivity of free-field voice testing to identify such an impairment because of an inability to hear a whispered voice at two feet (60 cm.) was 86 per cent or better with the specificity being in the region of 90 per cent. In the Rinne test the 256 Hz fork was superior to the 512 Hz fork (p less than 0.05) and the loudness comparison method superior to the threshold decay method (p less than 0.01) in detecting an air-bone gap. Combining the responses to the two forks did not improve the results. The Rinne test with the 256 Hz fork will identify correctly 48 per cent of individuals with a 15 dB, 69 per cent with a 20 dB, 87 per cent with a 25 dB, and 95 per cent with a 30 dB conductive impairment. In all instances the specificity is greater than 90 per cent.