The immune reconstitution inflammatory syndrome (IRIS) has emerged as an important early complication of antiretroviral therapy (ART) in resource-limited settings, especially in patients with tuberculosis. However, there are no consensus case definitions for IRIS or tuberculosis-associated IRIS. Moreover, previously proposed case definitions are not readily applicable in settings where laboratory resources are limited. As a result, existing studies on tuberculosis-associated IRIS have used a variety of non-standardised general case definitions. To rectify this problem, around 100 researchers, including microbiologists, immunologists, clinicians, epidemiologists, clinical trialists, and public-health specialists from 16 countries met in Kampala, Uganda, in November, 2006. At this meeting, consensus case definitions for paradoxical tuberculosis-associated IRIS, ART-associated tuberculosis, and unmasking tuberculosis-associated IRIS were derived, which can be used in high-income and resource-limited settings. It is envisaged that these definitions could be used by clinicians and researchers in a variety of settings to promote standardisation and comparability of data.