Résumé : Objective. Our study reports the results of the MIRA (MabThera In Rheumatoid Arthritis) registry, set up to collect data about clinical usage, patient profile, and retention of rituximab (RTX) treatment in daily clinical practice in Belgium. Methods. Patients with active rheumatoid arthritis (RA) who failed at least 1 anti-tumor necrosis factor (anti-TNF) treatment were included in our study between November 2006 and October 2011. At baseline, demographics, medication, disease history, disease activity, rheumatoid factor (RF), and anticyclic citrullinated peptide antibodies (anti-CCP) status were recorded. Evolution of the 28-joint Disease Activity Score (DAS28)-erythrocyte sedimentation rate, retreatments, and reasons for therapy stop were followed prospectively. Results. The MIRA registry included 649 patients, with mean disease duration of 12.8 ± 0.4 years and DAS28 values at inclusion of 5.85 ± 0.48. Patients received on average 2.82 ± 0.07 (range 1-9) RTX treatments, over a mean followup period of 93.1 ± 2.6 weeks. At database lock, 433 patients (66.7%) were still under RTX treatment, 182 (28.0%) had stopped treatment, and 34 (5.2%) were lost to followup. Ineffectiveness (n = 108, 59%) and safety concerns (n = 39, 22%) were the most frequent reasons for discontinuing RTX therapy. From 2006 to 2011, RTX practice patterns clearly evolved toward RTX being started in patients with a lower number of previously failed anti-TNF drugs and lower baseline DAS28 values. A lower number of previous anti-TNF drugs, and positivity for RF and anti-CCP, predicted more successful longterm treatment. RTX treatment provided adequate longterm disease control. Conclusion. In our daily practice study, RTX provided good longterm disease control and treatment retention in refractory patients with RA. Over the years, rheumatologists tended to start this treatment in patients with fewer previous anti-TNF treatments and lower disease activity. The Journal of Rheumatology Copyright © 2014. All rights reserved.