Background There is scant evidence to support target drug exposures for optimal tuberculosis outcomes. We therefore assessed whether pharmacokinetic/pharmacodynamic (PK/PD) parameters could predict 2-month culture conversion. Methods One hundred patients with pulmonary tuberculosis (65% HIV-co-infected) were intensively sampled to determine rifampicin, isoniazid and pyrazinamide plasma concentrations after 7-8 weeks of therapy, and pharmacokinetic parameters determined using non-linear-mixed-effects models. Detailed clinical data and sputum for culture were collected at baseline, 2 and 5-6 months. Minimum inhibitory concentrations (MIC) were determined on baseline isolates. Multivariate logistic regression and the assumption-free multivariate adaptive regression splines (MARS) were used to identify clinical and PK/PD predictors of 2-month culture conversion. Potential PK/PD predictors included 24-hour-area-under-the-curve (AUC0-24), peak concentration (Cmax), AUC0-24/MIC, Cmax/MIC and % time that concentrations persisted above MIC (%TMIC). Results 26% of patients had Cmax (mg/L) of rifampicin4.6 mg/L, higher isoniazid exposures were associated with improved rates of culture conversion. Conclusions PK/PD analyses using MARS identified isoniazid Cmax and rifampicin Cmax/MIC thresholds below which there is concentration-dependent antagonism that reduces 2-month sputum culture conversion.