BACKGROUND: Shorter treatment regimens for tuberculosis (TB) are deemed vital for advancing TB control. Murine studies have suggested potential new regimens; however, Phase II human studies of these drug combinations have not shown clear improvement in 2-month culture conversion over current therapy. Nevertheless, drugs such as rifapentine (RPT) may have additional sterilizing effects after 2 months that are difficult to measure in current Phase II studies. OBJECTIVES: To model potential bactericidal effects of RPT in a Phase III trial of a 4-month anti-tuberculosis regimen. METHODS: We developed a Markov model of anti-tuberculosis treatment to compare two regimens for treating TB: a 6-month standard (rifampin-based) treatment and a 4-month regimen using high-dose RPT. The primary outcome was the number of relapses. RESULTS: In the base-case scenario, standard therapy resulted in fewer relapses; improvement in 2-month culture conversion rates in the RPT arm did not change this result. However, while RPT has better sterilizing ability during months 3 and 4 (as observed in the mouse model), the 4-month regimen results in fewer relapses. CONCLUSIONS: Higher 2-month culture conversion rates are neither sufficient nor necessary for making a theoretical 4-month anti-tuberculosis treatment regimen advantageous.