The bleach digestion of sputum may improve the yield of smear microscopy but has not been validated in patients with HIV. Therefore we assessed the performance of bleach-digested smear microscopy among patients with HIV. One thousand three hundred and twenty one patients with chronic cough submitted three sputum samples for direct smear microscopy and were offered HIV tests. One sample was selected for a bleach-digested smear and another one was cultured. Patients were classified as having 'definite' (>or=2 positive smears), 'very likely' (smear-negative, culture- positive), 'less likely' (one smear-positive, culture-negative) and 'unlikely' (smear and culture negative) tuberculosis (TB). In all, 566/1045 (54%) patients were HIV positive and 731/1186 (62%) were culture positive. The digested smears were positive in 123/125 (98%) 'definite', 4/118 (3%) 'very likely' and 1/174 'unlikely' TB patients with HIV and in 125/127 (98%) 'definite', 2/74 (3%) 'very likely', 4/4 'less likely' and 2/127 'unlikely' TB without HIV. Three direct smears identified 252 (57%) and one digested smear 254 (57%) of the 444 patients with 'definite' or 'very likely' TB. One bleach-digested smear performed similarly to three direct smears. Both methods were less sensitive in HIV-positive patients. Further studies are needed to compare the performance of the two methods under operational conditions.