The aim of this study was to compare the performance of the T-SPOT.TB test, a T-cell-based test, with the tuberculin skin test (TST) in the diagnosis of latent tuberculosis (TB) infection. The study was carried out in 138 immunosuppressed haematology patients who had been nosocomially exposed to a case of smear-positive TB. Overall, 44.2% of the contacts were positive by T-SPOT.TB test, and 17.4% by TST (concordance 67.8%). The apparent prevalence of infection fell from 25.9 to 14.5% with the TST with increasing immunosuppression, although this difference was not significant. In contrast, the apparent prevalence of infection with the T-SPOT.TB test was unaffected at 44.6 and 44.3%, respectively. The T-SPOT.TB test had an overall indeterminate rate of 4.3%, and this was also unaffected by the level of immunosuppression. This study suggests that the T-SPOT.TB test maintains its sensitivity and performance in immunocompromised patients, identifying a large number of truly infected patients anergic to the tuberculin skin test.