A total of 822 patients with newly diagnosed pulmonary tuberculosis were assigned randomly to one of 3 daily rifampin-isoniazid (RIF-INH) regimens: 450, 600, or 750 mg of RIF in combination with 300 mg of INH. After an initial 20 weeks of therapy with RIF-INH, patients recieved 300 mg of INH and 15 mg of ethambutol (EMB) per kg of body weight for either 12 or 18 months after their sputum cultures became negative. The rate of bacteriologic conversion of sputum among the 3 RIF-INH regimens was compared for 552 patients who completed the 20 weeks of RIF-INH therapy. Apporximately 60 per cent of these patients also completed their assigned INH-EMB therapy and were examined for relapse for at least one year after therapy was stopped. There was no significant difference in the rate of sputum conversion or rate of relapse between the group of patients who received 600 mg of RIF and those who received 750 mg of RIF. However, the 450-mg RIF regimen was significantly less effective than the other 2 regimens, as manifested by a lower rate of sputum conversion and a higher rate of treatment failures. Further analysis showed that RIF dosages of less than 9 mg per kg of body weight per day may be inadequate for treatment of pulmonary tuberculosis. The acceptability of these regimens was high, and the incidence of adverse reactions requiring discontinuation of RIF-INH therapy was quite low (3.3 per cent). A large proportion of patients (44 per cent) developed increased concentrations of transaminase during therapy with RIF-INH. These abnormalities were usually transient and, in most cases, of no clinical significance. In the relapse analysis, 12 months of chemotherapy after sputum conversion was shown to be as effective as 18 months of therapy after conversion of these RIF-containing regimens.