Six-month regimens that include rifampin for the treatment of tuberculosis in patients without human immunodeficiency virus (HIV) infection are recommended because of low percentage of relapses. Whether a similar duration of therapy should be used to treat tuberculosis in HIV-infected patients is unclear. Six studies of patients with HIV-infection and 3 of patients without HIV infection were reviewed and compared. The studies differed in terms of design, eligibility criteria, site of disease, frequency of dosing, dose administration methods, and outcome definitions. Among HIV-infected patients, the following percentages were found: cure, 59.4%--97.1%; treatment success, 34.0%--100%; effective treatment, 29.4%--88.2%; and relapse, 0%--10%. In those without HIV infection, percentages were as follows: cure, 62.3%--88.0%; treatment success, 91.2%--98.8%; effective treatment, 70.6%--83.8%; and relapse, 0%--3.4%. Although the rate of relapse appeared to be higher in some studies of HIV-infected patients with tuberculosis, this review demonstrates the limitation in the use of relapse as the exclusive outcome of interest when comparing studies.