SETTING: Programmatic implementation of decentralized rapid drug susceptibility testing (DST) in Lima, Peru. OBJECTIVE: Pre-post analysis compared time to diagnosis, treatment outcome and survival among patients tested with direct nitrate reductase assay (NRA) vs. indirect conventional methods. DESIGN: From 2005 to 2009, we prospectively followed all patients referred for DST before (control) and after (intervention) NRA implementation. Among those referred for DST, NRA was used for smear-positive samples of patients with no prior history of multidrug resistance or treatment for multidrug-resistant tuberculosis (TB). Data were abstracted from patient charts and laboratory registers. Endpoints were favorable outcomes, time to result and time to death. RESULTS: Of those patients who met the criteria for NRA, 740 underwent NRA and 621 underwent conventional DST. NRA yielded test results for 78.4% of cases vs. 68.8% for conventional DST (P < 0.0001); the median time to result was 44 vs. 133 days, respectively (adjusted HR 0.64, 95%CI 0.56-0.73). Among individuals without previous anti-tuberculosis treatment, NRA was associated with a favorable treatment outcome (adjusted OR 1.39, 95%CI 1.01-1.90) and prolonged survival (adjusted HR 0.53, 95%CI 0.31-0.90). CONCLUSION: Direct NRA significantly shortened time to test result and improved treatment outcomes and survival in certain groups. © 2012 The Union.