The targets for tuberculosis control, framed within the United Nations' Millennium Development Goals, are to ensure that the incidence per head of tuberculosis is falling by 2015, and that the 1990 prevalence and mortality per head are halved by 2015. In monitoring progress in tuberculosis control, the ultimate aim for all countries is to count tuberculosis cases (incidence) accurately through routine surveillance. Disease prevalence surveys are costly and laborious, but give unbiased measures of tuberculosis burden and trends, and are justified in high-burden countries where many cases and deaths are missed by surveillance systems. Most countries in which tuberculosis is highly endemic do not yet have reliable death registration systems. Verbal autopsy, used in cause-of-death surveys, is an alternative, interim method of assessing tuberculosis mortality, but needs further validation. Although several new assays for Mycobacterium tuberculosis infection have recently been devised, the tuberculin skin test remains the only practical method of measuring infection in populations. However, this test typically has low specificity and is therefore best used comparatively to assess geographical and temporal variation in risk of infection. By 2015, every country should be able to assess progress in tuberculosis control by estimating the time trend in incidence, and the magnitude of reductions in either prevalence or deaths. © 2008 Elsevier Ltd. All rights reserved.