A cohort of rheumatoid arthritis (RA) patients in the Lombardy Rheumatology Network (LOHREN) registry and receiving anti-TNF therapy was evaluated after 6, 12, 24, and 36 months. Of the 1114 patients in the registry 1064 met the clinical criteria for inclusion with 519 receiving infliximab, 303 adalimumab, and 242 etanercept. The therapeutic survival curve of these patients showed that the likelihood of continuing anti-TNF therapy was 78.8% after 12 months, 65.2% after 24 months, and 52.9% after 36 months, with a risk of dropout similar for inefficacy and adverse events. There were 405 anti-TNF therapy discontinuations (38.1%): 180 (16.9%) due to inefficacy, 194 (18.2%) adverse events, and 31 (2.9%) other reasons. Four deaths (2 septicemia, 1 postinfective cerebritis, 1 heart failure) were considered to be related to anti-TNF therapy. Of the discontinuations, 219 (54.1%) occurred within the first 12 months: 110 due to adverse events, 89 inefficacy, and 20 due to other reasons. After 36 months, the likelihood of survival on etanercept (62.5%) was significantly greater than the likelihood of survival on infliximab (49.1%) or adalimumab (53.6%). A higher risk of therapy discontinuations due to adverse events was associated with increasing age, a corticosteroid > 5 mg/day, a high erythrocyte sedimentation rate (ESR), a higher risk of therapy discontinuations due to inefficacy was associated with the previous use of > or = 4 disease-modifying antirheumatic drugs (DMARDs) and a high ESR. Comorbidities, increasing DAS28 values and co-therapy with methotrexate were associated with a lower risk of discontinuation.