Rationale: We examined whether increased rifampin doses could shorten standard therapy for tuberculosis without increased toxicity. Objectives: To assess the differences across three daily oral doses of rifampin in change in elimination rate of Mycobacterium tuberculosis in sputum and frequency of rifampin-related adverse events. Methods: We conducted a blinded, randomized, controlled phase 2 clinical trial of 180 adults with new smear-positive pulmonary tuberculosis, susceptible to isoniazid and rifampin. We randomized 1:1:1 to rifampin at 10, 15, and 20 mg/kg/d during the intensive phase. We report the primary efficacy and safety endpoints: change in elimination rate of M. tuberculosis log10 colony-forming units and frequency of grade 2 or higher rifampin-related adverse events. We report efficacy by treatment arm and by primary (area under the plasma concentration–time curve [AUC]/minimum inhibitory concentration [MIC]) and secondary (AUC) pharmacokinetic exposure. Measurements and Main Results: Each 5-mg/kg/d increase in rifampin dose resulted in differences of 20.011 (95% confidence interval, 20.025 to 10.002; P = 0.230) and 20.022 (95% confidence interval, 20.046 to 20.002; P = 0.022) log10 cfu/ml/d in the modified intention-to-treat and per-protocol analyses, respectively. The elimination rate in the per-protocol population increased significantly with rifampin AUC0–6 (P = 0.011) but not with AUC0–6/MIC99.9 (P = 0.053). Grade 2 or higher rifampin-related adverse events occurred with similar frequency across the three treatment arms: 26, 31, and 23 participants (43.3%, 51.7%, and 38.3%, respectively) had at least one event (P = 0.7092) up to 4 weeks after the intensive phase. Treatment failed or disease recurred in 11 participants (6.1%). Conclusions: Our findings of more rapid sputum sterilization and similar toxicity with higher rifampin doses support investigation of increased rifampin doses to shorten tuberculosis treatment. Clinical trial registered with www.clinicaltrials.gov (NCT 01408914).