Five methods of identifying tuberculosis suspects were investigated in the Machakos District of Kenya by: (1) 3-monthly interrogation of the Community Elders, (2) interrogation of household heads, (3) identifying suspects amongst outpatients attending local health units, (4) examination of patients registered during the previous 10 years in the District Tuberculosis Register and also (5) their close contacts. Sputum was bacteriologically examined by smear and culture from suspects found by all the methods. The initial interrogation of the Elders yielded 216 suspects, of whom 9 were culture-positive, including 6 smear-positive. Reinterrogating the Elders 4 times at 3-monthly intervals produced a further 114 suspects including 4 culture-positive cases (3 being smear-positive). The examination of a second sputum specimen from suspects after a 3-month interval yielded 4 further culture-positive cases (all smear-negative) but the examination of a third specimen after a further 3 months yielded no further cases. A single interrogation of 1093 household head suspects yielded 22 culture-positive cases, including 11 smear-positive. The response in 5 health units covering a population of about 24 500 was poor. During a 2-year period only 109 suspects were recorded; 7 were culture-positive, including 3 smear-positive. Of 61 cases of tuberculosis registered during the previous 10 years, 8 were currently culture-positive, 5 being smear-positive. Of 318 household contacts of these cases, 6 were culture-positive cases, 2 being smear-positive. The problems presented by different active case-finding methods are discussed, identifying those that appear promising and those unpromising.