A randomized comparison has been made of three times weekly rifampin plus isoniazid (HR3) with rifapentine plus isoniazid given once weekly (HRp1) or on 2 of 3 wk (HRp1.2/3) in the continuation phase of 6-mo regimens (each starting with an initial 2 mo of 4-drug therapy) for the treatment of pulmonary tuberculosis in 672 Chinese patients in Hong Kong. Because of poor bioavailability of the rifapentine used (produced in China), its dose size was increased from 600 mg initially to about 750 mg in the last third of patients to obtain serum concentrations similar to those with rifapentine of Western origin; all doses were given after a meal promoting absorption. After initial exclusions, an intent to treat analysis, done on the remaining 592 patients, showed 45 adverse treatment events in 7 of 190 HR3 patients, in 17 of 199 HRp1 patients, and in 21 of 203 HRp1.2/3 patients; of these, 42 were bacteriological or radiographic relapses after the end of treatment (HR3 versus HRp1, p = 0.04; HR3 versus HRp1.2/3, p = 0.01). Patients with organisms initially sensitive or resistant to isoniazid or streptomycin had similar relapse rates. The high relapse rate in the HRp1 regimen suggests that the rifapentine dose should be increased. Similarity of relapse rates, 8.9% and 10.4%, after the HRp1 and HRp1.2/3 regimens, respectively, indicates that irregularity in taking rifapentine/isoniazid could be tolerated. The few adverse side effects in the continuation phase in the rifapentine regimens were less frequent than in the HR3 regimen.