Background: Immunocompromised persons infected with Mycobacterium tuberculosis (MTB) have increased risk of tuberculosis (TB) reactivation, but their management is hampered by the occurrence of false-negative results of the tuberculin skin test (TST). The T-cell interferon (IFN)-γ release blood assays T-SPOT.TB (TS.TB) [Oxford Immunotec; Abingdon, UK] and QuantiFERON-TB Gold In-Tube (QFT-IT) [Cellestis Ltd; Carnegie, VIC, Australia] might improve diagnostic accuracy for latent TB infection (LTBI) in high-risk persons, although their performance in different groups of immunocompromised patients is largely unknown. Methods and results: Over a 1-year period, we prospectively enrolled patients in three different immunosuppressed groups, as follows: 120 liver transplantation candidates (LTCs); 116 chronically HIV-infected persons; and 95 patients with hematologic malignancies (HMs). TST, TS.TB, and QFT-IT were simultaneously performed, their results were compared, and intertest agreement was evaluated. Overall, TST provided fewer positive results (10.9%) than TS.TB (18.4%; p < 0.001) and QFT-IT (15.1%; p = 0.033). Significantly fewer HIV-infected individuals had at least one positive test (9.5%) compared with LTCs (35.8%; p < 0.001) and patients with HMs (29.5%; p < 0.001). Diagnostic agreement between tests was moderate (κ = 0.40 to 0.65) and decreased in the HIV-infected group when the results of the TS.TB were compared with either TST (κ = 0.16) or QFT-IT (κ = 0.19). Indeterminate blood test results due to low positive control values were significantly more frequent with QFT-IT (7.2%) than with TS.TB (0.6%; p < 0.001). Conclusions: Blood tests identified significantly more patients as being infected with MTB than TST, although diagnostic agreement varied across groups. Based on these results, we recommend tailoring application of the new blood IFN-γ assays for LTBI in different high-risk groups and advise caution in their current use in immunosuppressed patients. Copyright © 2009 American College of Chest Physicians.