Tapering of disease-modifying antirheumatic drugs (DMARDs) to drug-free remission is an attractive treatment goal for patients with rheumatoid arthritis, although long-term effects of tapering and withdrawal remain unclear. We compared 3-year risks of flare between three conventional synthetic DMARD treatment strategies in patients with rheumatoid arthritis in sustained remission. In this open-label, randomised controlled, non-inferiority trial, we enrolled patients aged 18-80 years with rheumatoid arthritis who had been in sustained remission for at least 1 year on stable conventional synthetic DMARD therapy. Patients from ten hospitals in Norway were randomly assigned (2:1:1) with centre stratification to receive stable conventional synthetic DMARDs, half-dose conventional synthetic DMARDs, or half-dose conventional synthetic DMARDs for 1 year followed by withdrawal of all conventional synthetic DMARDs. The primary endpoint of this part of the study was disease flare over 3 years, analysed as flare-free survival and risk difference in the per-protocol population with a non-inferiority margin of 20%. This trial is registered with ClinicalTrials.gov (NCT01881308) and is completed. Between June 17, 2013, and June 18, 2018, 160 patients were enrolled and randomly assigned to receive stable-dose conventional synthetic DMARDs (n=80), half-dose conventional synthetic DMARDs (n=42), or half-dose conventional synthetic DMARDs tapering to withdrawal (n=38). Four patients did not receive the intervention and 156 patients received the allocated treatment strategy. One patient was excluded due to major protocol violation and 155 patients were included in the per-protocol analysis. 104 (67%) of 156 patients were women and 52 (33%) were men. 139 patients completed 3-years follow-up without major protocol violation; 68 (87%) of 78 patients in the stable-dose group, 36 (88%) of 41 patients in the half-dose group and 35 (95%) of 37 patients in the half-dose tapering to withdrawal group. During the 3-year study period, 80% (95% CI 69-88%) were flare-free in the stable-dose group, compared with 57% (41-71%) in the half-dose group and 38% (22-53%) in the half-dose tapering to withdrawal group. Compared with stable-dose conventional synthetic DMARDs, the risk difference of flare was 23% (95% CI 6-41%, p=0·010) in the half-dose group and 40% (22-58%, p<0·0001) in the half-dose tapering to withdrawal group, non-inferiority was therefore not shown. Adverse events were reported in 65 (83%) of 78 patients in the stable-dose group, 36 (90%) of 40 patients in the half-dose group, and 36 (97%) of 37 patients in the half-dose tapering to withdrawal group. One death occurred in the stable-dose conventional synthetic DMARD group (sudden death considered unlikely related to the study medication). Two conventional synthetic DMARD tapering strategies were associated with significantly lower rates of flare-free survival compared with stable conventional synthetic DMARD treatment, and the data do not support non-inferiority. However, drug-free remission was achiveable for a significant subgroup of patients. This trial provides information on risk and benefits of different treatment strategies important for shared decision making. Research Council of Norway and South-Eastern Norway Regional Health Authority.