In rheumatic patients candidate to anti-TNF-alpha treatment, there is an increased risk of developing tuberculosis (TB). The tuberculin skin test (TST), the standard diagnostic test for latent tuberculosis infection (LTBI), suffers low specificity and sensitivity. Here, we compared the performance characteristics of an in-house ELISPOT-IFN-gamma assay (using a restricted pool of Mycobacterium tuberculosis-specific peptides or MTP) to TST for the diagnosis of LTBI in 69 rheumatic patients candidate to anti-TNF-alpha treatment and in 60 healthy LTBI individuals. Among the 69 patients enrolled, 17 (25%) had a positive TST response and 15 (22%) a positive ELISPOT-MTP response. Among the patients with a positive TST result, eight had a positive and nine a negative ELISPOT-MTP response, whereas among the 49 patients with a negative TST result, 42 were ELISPOT-MTP negative, but seven (14%) were ELISPOT-MTP positive, with three indeterminate results. The agreement between the two tests was poor (k = 0.341, 95% CI = 0.060 to 0.622) and the test of symmetry was not significant (P = 0.8). Considering the ELISPOT assay, rheumatic patients had a reduced number of spot-forming cells after stimulation of lymphocytes with PHA or PPD when compared with healthy LTBI individuals. Thus, the ELISPOT-IFN-gamma assay performs better than the TST in recognizing patients with LTBI, on one hand reducing the number of patients submitted to isoniazid prophylaxis, and on the other hand, since the assay is less biased by immunosuppressive regimens than TST, recognizing LTBI patients among those with a negative TST response.