The clinical utility of therapeutic drug monitoring in tuberculosis has not been adequately evaluated by controlled clinical trials. To examine the relationship between slow culture conversion and peak plasma rifampicin level (Cmax-rfm) in a case-control study, patients with persistence of positive sputum smear despite at least 8 weeks of directly observed treatment with standard pyrazinamide-containing regimens were enrolled prospectively in government chest clinics from 16 December 2005 to 15 November 2006. Patients with multidrug-resistant tuberculosis, human immunodeficiency virus infection, or poor treatment adherence were excluded. Cases referred to patients with persistence of positive culture whereas controls had negative culture despite positive smear. Blood was checked at 2 and 4 hours post-dosing to capture Cmax-rfm. A cohort of 88 patients was identified. After excluding 16 patients, there were 36 controls and 36 cases. None had symptoms of malabsorption. Cmax-rfm was below 6 mg/l among 47% of controls and 44% of cases. Univariate and multiple logistic regression analyses showed no significant association between slow culture conversion and Cmax-rfm after logarithmic transformation. Thus, there is probably no association between Cmax-rfm and slow culture conversion.