BACKGROUND: Four months of rifampin treatment for latent tuberculosis infection is safer, has superior treatment completion rates, and is as effective as 9 months of isoniazid. However, daily medication costs are higher for a 4-month rifampin regimen than a 9-month isoniazid regimen. OBJECTIVE: To compare health care use and associated costs of 4 months of rifampin and 9 months of isoniazid. DESIGN: Health system cost comparison using all health care activities recorded during 2 randomized clinical trials. (ClinicalTrials.gov: NCT00931736 and NCT00170209). SETTING: High-income countries (Australia, Canada, Saudi Arabia, and South Korea), middle-income countries (Brazil and Indonesia), and African countries (Benin, Ghana, and Guinea). PARTICIPANTS: Adults and children with clinical or epidemiologic factors associated with increased risk for developing tuberculosis that warranted treatment for latent tuberculosis infection. MEASUREMENTS: Health system costs per participant. RESULTS: A total of 6012 adults and 829 children were included. In both adults and children, greater health system use and higher costs were observed with 9 months of isoniazid than with 4 months of rifampin. In adults, the ratios of costs of 4 months of rifampin versus 9 months of isoniazid were 0.76 (95% CI, 0.70 to 0.82) in high-income countries, 0.90 (CI, 0.85 to 0.96) in middle-income countries, and 0.80 (CI, 0.78 to 0.81) in African countries. Similar findings were observed in the pediatric population. LIMITATION: Costs may have been overestimated because the trial protocol required a minimum number of follow-up visits, although fewer than recommended by many authoritative guidelines. CONCLUSION: A 4-month rifampin regimen was safer and less expensive than 9 months of isoniazid in all settings. This regimen could be adopted by tuberculosis programs in many countries as first-line therapy for latent tuberculosis infection. PRIMARY FUNDING SOURCE: Canadian Institutes of Health Research.