A systematic review was conducted on articles reporting tuberculin surveys to estimate the prevalence and annual risk of tuberculosis infection (ARI) in indigenous peoples of the Americas. MEDLINE and LILACS databases were searched using the following combination of descriptors: "tuberculosis", "South American Indians", "North American Indians", "tuberculin skin test", "PPD", "risk assessment", and "BCG vaccine". Longitudinal and cross-sectional studies published from 1974 to 2004 were selected if the outcome was Mycobacterium tuberculosis infection defined by tuberculin reactions > 5mm or > 10mm. From a total of 54 candidate studies generated with the descriptors, 16 were selected for review. Altogether, they comprised 13,186 individuals, 50% of whom were below 15 years of age. Estimates of prevalence of tuberculosis infection ranged from 0% to 93.4%, whereas the ARI ranged from 0% to 35.4%. Analysis of heterogeneity in environmental conditions and methods of infection assessment in the studies showed that the tuberculin test is useful for clinical decisions (e.g., to initiate chemoprophylaxis), and that tuberculin surveys are useful to assess tuberculosis control measures among indigenous peoples.