Tuberculosis occurring with human immunodeficiency virus (HIV) infection is a serious and growing public health problem. We have carried out a randomised clinical trial of a 12-month course of isoniazid plus vitamin B6 versus vitamin B6 alone in Port-au-Prince, Haiti, to assess the efficacy of isoniazid in preventing active tuberculosis in symptom-free HIV-infected individuals. The effect of prophylaxis on the development of HIV disease, AIDS, and death was also investigated. 118 subjects were assigned treatment with isoniazid plus B6 (n = 58) or B6 alone (n = 60) between 1986 and 1989. The treatment groups were similar at study entry in demographic, clinical, and immunological characteristics. Interim analysis in 1990 revealed no significant difference in tuberculosis outcome measures. Follow-up was continued until 1992, at which time significant protection by isoniazid against the development of tuberculosis was apparent, both for the whole study population and for subjects positive for purified protein derivative of tuberculin (PPD). The incidence of tuberculosis was lower in isoniazid recipients than in patients who received B6 alone (2.2 vs 7.5 per 100 person-years). The relative risk of tuberculosis was 3.4 (95% CI 1.1-10.6) for B6 alone versus isoniazid plus B6 (p < 0.05). Isoniazid also delayed progression to HIV disease and AIDS and death. Thus isoniazid effectively decreases the incidence of tuberculosis and delays the onset of HIV-related disease in symptom-free HIV-seropositive individuals. Isoniazid prophylaxis should be considered for HIV-seropositive, PPD-positive subjects, and may also be appropriate for PPD-negative patients in areas where tuberculosis is highly endemic.