PA-824 is in phase II clinical testing to treat tuberculosis. At a dose of 100 mg/kg of body weight, it has demonstrated bactericidal activity during the initial and continuation phases of treatment in a murine model of tuberculosis. In a prior study, substitution of PA-824 for isoniazid in the first-line regimen of rifampin, isoniazid, and pyrazinamide resulted in significantly lower CFU counts at 2 months and shorter time to culture-negative conversion. However, the study design prevented a rigorous assessment of the relapse rate after completion of therapy. The current experiment was designed to assess (i) the extent of the beneficial effect of substituting PA-824 for isoniazid in the first-line regimen, (ii) the influence of the PA-824 dose on the same effect, and (iii) the activity of each one-, two-, and three-drug combination of rifampin, PA-824, and pyrazinamide. Mice were infected by the aerosol route and initiated on treatment 14 days later with more than 7 log(10) CFU per lung. Treatment with rifampin and pyrazinamide was more effective than treatment with rifampin, isoniazid, and pyrazinamide at reducing the lung CFU count, consistent with past evidence of isoniazid's antagonism in this model. The addition of PA-824 at 12.5 and 25 mg/kg/day did not increase the activity of rifampin plus pyrazinamide, but the addition of PA-824 at 50 and 100 mg/kg/day did increase the activity in a dose-dependent manner. The combination of rifampin, PA-824 (100 mg/kg), and pyrazinamide rendered all mice culture negative after 2 months of treatment and free of relapse after 4 months of treatment, while some mice receiving rifampin, isoniazid, and pyrazinamide remained culture positive and 15% relapsed after completing 4 months of treatment. The two-drug combination of PA-824 and pyrazinamide displayed synergistic activity that was equivalent to that of the standard first-line regimen. Together, these results support the evaluation of regimens based on the combination of rifampin, PA-824, and pyrazinamide in phase II clinical trials while demonstrating several potential pitfalls in the evaluation of new drug combinations in a murine model of tuberculosis.