BACKGROUND We obtained estimates of the incidence of tuberculosis (TB) among patients receiving HAART and identified determinants of the incidence. METHODS We analyzed the incidence of TB during the first 3 years after initiation of HAART among 17,142 treatment-naive, AIDS-free persons starting HAART who were enrolled in 12 cohorts from Europe and North America. We used univariable and multivariable Poisson regression models to identify factors associated with the incidence. RESULTS During the first 3 years (36,906 person-years), 173 patients developed TB (incidence, 4.69 cases per 1000 person-years). In multivariable analysis, the incidence rate was lower for men who have sex with men, compared with injection drug users (relative rate, 2.46; 95% confidence interval [CI], 1.51-4.01), heterosexuals (relative rate, 2.42; 95% CI, 1.64-3.59), those with other suspected modes of transmission (relative rate, 1.66; 95% CI, 0.91-3.06), and those with a higher CD4+ count at the time of HAART initiation (relative rate per log2 cells/microL, 0.87; 95% CI, 0.84-0.91). During 28,846 person-years of follow-up after the first 6 months of HAART, 88 patients developed TB (incidence, 3.1 cases per 1000 person-years of follow-up). In multivariable analyses, a low baseline CD4+ count (relative rate per log2 cells/microL, 0.89; 95% CI, 0.83-0.96), 6-month CD4+ count (relative rate per log2 cells/microL, 0.90; 95% CI, 0.81-0.99), and a 6-month HIV RNA level >400 copies/mL (relative rate, 2.21; 95% CI, 1.33-3.67) were significantly associated with the risk of acquiring TB after 6 months of HAART. CONCLUSIONS The level of immunodeficiency at which HAART is initiated and the response to HAART are important determinants of the risk of TB. However, this risk remains appreciable even among those with a good response to HAART, suggesting that other interventions may be needed to control the TB epidemic in the HIV-infected population.