Tuberculosis (TB) remains a global health problem, and South Africa (SA) has one of the world's worst TB epidemics. The World Health Organization (WHO) estimated in 1999 that one-third of the world's population was latently infected with TB. In SA up to 88% of HIV-uninfected young adults (31 - 35 years) are latently infected with TB. In the most recent meta-analysis, 6 - 12 months of isoniazid preventive therapy (IPT) was associated with a lower incidence of active TB than placebo (relative risk (RR) 0.68; 95% confidence interval (CI) 0.54 - 0.85), with the greatest benefit among individuals with a positive tuberculin skin test (TST) (RR 0.38; 95% CI 0.25 - 0.57). A clinical trial of IPT given with antiretroviral therapy (ART) for 12 months reduced TB incidence by 37% compared with ART alone (hazard ratio (HR) 0.63; 95% CI 0.41 - 0.94). The effect of IPT is limited in high-burden countries. IPT for 36 months v. 6 months reduced TB incidence among HIV-positive, TST-positive participants by 74% (HR 0.26; 95% CI 0.09 - 0.80). A study of more than 24 000 goldminers confirmed that IPT is safe, with only 0.5% experiencing adverse events. A meta-analysis of studies of IPT since 1951 did not show an increased risk of developing resistance. Alternative TB preventive therapy regimens, including high-dose isoniazid and rifapentine given weekly for 3 months, have been shown to have similar efficacy to IPT. Mathematical modelling suggests that scaling up continuous IPT targeted to HIV-positive persons, when used in combination with other treatment and prevention strategies, may substantially improve TB control.