The World Health Organization estimates that a third of the world's population is infected with Mycobacterium tuberculosis. Every second, one person becomes newly infected with tuberculosis (TB). In the past two decades, the spread of human immunodeficiency virus infection, worsening poverty and deteriorating health services have resulted in a steady increase in the overall incidence of TB globally. With treatment of latent TB infection (LTBI), the number of infected persons who develop active TB can be significantly diminished. Prevention through treatment of LTBI should therefore be an integral part of the control of TB. Although only a minority of those with LTBI will develop active disease, the risk varies substantially according to the time since infection and medical risk factors. If persons at low risk for TB are selected for preventive chemotherapy, the individual and public health benefits are low, and a large number will have to be treated to prevent a single active case. It is therefore important to identify and treat patients who are at high risk of disease. Tools for rapid and reliable identification of persons with LTBI who are most likely to progress to active disease are urgently needed, as this will permit rational use of preventive treatment by restricting treatment to those patients with the most favourable risk/benefit ratio. The major challenges are efficient identification of those at highest risk of developing disease and ensuring treatment completion with a non-toxic regimen. If these can be overcome, preventive treatment holds the promise to substantially assist in the achievement of global control of TB.