In a study designed to compare the Mantoux with the tine and 'Imotest' multiple-puncture tuberculin tests 200 subjects underwent a Mantoux test (10 IU PPD-Weybridge) and, simultaneously, either a tine test or an imotest. All tests were read after 72 h by two observers independently. The false-negative rate for the Mantoux was estimated by immediate repetition of the Mantoux in subjects with an initially negative Mantoux but positive multiple-puncture test. When the criterion for a positive reaction was at least 5 mm of induration for the Mantoux and at least 2 mm of induration for the multiple-puncture tests, the false-negative rate of 27% for the imotest was significantly higher than the 4% for the tine and 5% for the Mantoux (p less than 0.001 in each case). Positive multiple-puncture tests were always associated with a Mantoux test that was positive on either the first or second application. Observers agreed in their interpretation of 98% of all tests, and there were no significant differences between testers in frequency of false-negative results for any test. The tine test, when carefully applied and correctly interpreted, gives results which correlate well with those given by the standard Mantoux test. The imotest is less reliable and has no advantages over the tine test.