SETTING: Clinical trial in 672 patients with newly diagnosed pulmonary tuberculosis in Hong Kong. After an initial 2 months of a four-drug intensive phase consisting of thrice-weekly streptomycin, isoniazid, rifampicin and pyrazinamide (SHRZ), a random allocation was made to a continuation phase of once-weekly 600 mg rifapentine + 15 mg/kg isoniazid (HRp1), HRp1 given in 2 of every 3 weeks (HRp1.2/3), or to thrice-weekly isoniazid + rifampicin (HR3), the standard treatment in Hong Kong. OBJECTIVE: Final report evaluating adverse events (46 relapses and one failure) after 5 years and the prognostic influence of various factors. METHODS: Kaplan-Meier analysis of adverse events and Cox proportional hazards analysis of prognostic factors. RESULTS: The two rifapentine regimens, HRp1 and HRp1.2/3 had similar final rates of adverse events of 10.8% and 11.7%, respectively, compared to 4.2% for the HR3 regimen (P = 0.02 and 0.009, respectively). In the initial univariate proportional hazards analysis, adverse events were significantly related to the regimen, age, sex, pretreatment radiographic extent of disease and cavitation, and also to sputum culture at 2 months. In the final multivariate analysis, after step-wise removal of non-significant factors, adverse events were related only to the regimen, patients' sex and pretreatment radiographic extent of disease. Elderly male patients were more at risk of an adverse event, as were those with more severe disease. Adverse events occurred at life table rates of 9.0% in patients with drug-sensitive strains and in 8.9% of those with initially isoniazid-resistant organisms at 5 years. CONCLUSIONS: The two rifapentine regimens were unsatisfactory because of their high incidence of adverse events. Isoniazid appeared not to contribute to preventing relapse. Further studies with increased rifapentine dosage are necessary.