Gamma interferon release assays (IGRAs) are increasingly used for latent Mycobacterium tuberculosis infection (LTBI) screening in patients with rheumatic diseases starting anti-tumor necrosis factor (anti-TNF) therapies. We compared the performances of two IGRAs, an enzyme-linked immunospot release assay (T-SPOT.TB) and an enzyme-linked immunosorbent assay (QuantiFERON-TB Gold In Tube [QFT-GIT]), to that of tuberculin skin testing (TST) for LTBI screening of 157 consecutive rheumatic patients starting anti-TNF therapies. Among 155 patients with valid results, 58 (37%) were positive by TST, 39 (25%) by T-SPOT.TB assay, and 32 (21%) by QFT-GIT assay. IGRAs were associated more strongly with at least one risk factor for tuberculosis (TB) than TST. Risk factors for a positive assay included chest X-ray findings of old TB (TST), advanced age (both IGRAs), origin from a country with a high TB prevalence, and a positive TST (T-SPOT.TB assay). Steroid use was negatively associated with a positive QFT-GIT assay. The agreement rate between IGRAs was 81% (kappa rate = 0.47), which was much higher than that observed between an IGRA and TST. If positivity by either TST or an IGRA was required for LTBI diagnosis, then the rate of LTBI would have been 46 to 47%, while if an IGRA was performed only for TST-positive patients, the respective rate would have been 11 to 17%. In conclusion, IGRAs appear to correlate better with TB risk than TST and should be included in TB screening of patients starting anti-TNF therapies. In view of the high risk of TB in these patients, a combination of one IGRA and TST is probably more appropriate for LTBI diagnosis.