A major challenge in tuberculosis control is the diagnosis and treatment of latent tuberculosis infection. Until recently, there were no alternatives to the tuberculin skin test (TST) for diagnosing latent tuberculosis. However, an alternative has now emerged in the form of a new in-vitro test: the interferon-gamma assay. We did a systematic review to assess the performance of interferon-gamma assays in the immunodiagnosis of tuberculosis. By searching databases, contacting experts and test manufacturers, we identified 75 relevant studies. The results suggest that interferon-gamma assays that use Mycobacterium tuberculosis-specific region of difference 1 (RD1) antigens (such as early secretory antigenic target 6 and culture filtrate protein 10) may have advantages over the TST, in terms of higher specificity, better correlation with exposure to M tuberculosis, and less cross-reactivity due to BCG vaccination and non-tuberculous mycobacterial infection. However, interferon-gamma assays that use RD1 antigens in isolation may maximise specificity at the cost of sensitivity. Assays that use cocktails of RD1 antigens seem to overcome this problem, and such assays have the highest accuracy. RD1-based interferon-gamma assays can potentially identify those with latent tuberculosis who are at high risk for developing active disease, but this requires confirmation. There is inadequate evidence on the value of interferon-gamma assays in the management of immunocompromised individuals, children, patients with extrapulmonary or non-tuberculous mycobacterial disease, and populations in countries where tuberculosis is endemic. Current evidence suggests that interferon-gamma assays based on cocktails of RD1 antigens have the potential to become useful diagnostic tools. Whether this potential can be realised in practice remains to be confirmed in well designed, long-term studies.