ObjectivesTo compare the effectiveness of longstanding (>52 weeks), supervised exercise therapy with usual care in adults with rheumatoid arthritis (RA) and severe functional limitations.MethodsParticipants were randomised 1:1 to the intervention (individualised goal-setting, active exercises, education and self-management regarding physical activity) or usual care. Primary endpoint was the change in the Patient-Specific Complaints activity ranked 1 (PSC1, 0–10) at 52 weeks. Secondary endpoints included the PSC activities ranked 2 and 3 (PSC2, PSC3), Health Assessment Questionnaire-Disability Index (HAQ-DI), Rheumatoid Arthritis Quality of Life Questionnaire (RAQoL), 6-minute walk test (6MWT), Patient Reported Outcome Measurement Information System Physical Function-10 (PROMIS PF-10) and the Short Form-36 Physical and Mental Component Summary Scales (SF-36 PCS and MCS). (Serious) Adverse events (AEs) were recorded. Measurements were done by blinded assessors. Analyses at 52 weeks were based on the intention-to-treat principle.ResultsIn total, 217 people (90% female, age 58.8 (SD 12.9) years) were randomised (n=104 intervention, n=98 usual care available for analyses). At 52 weeks, the improvement of the PSC1 was significantly larger in the intervention group (mean difference (95% CI) −1.7 (−2.4, –1.0)). Except for the SF-36 MCS, all secondary outcomes showed significantly greater improvements favouring the intervention (PSC2 −1.8 (−2.4, –1.1), PSC3 −1.7 (−2.4, −1.0), PROMIS PF-10 +3.09 (1.80, 4.38), HAQ-DI −0.17 (−0.29, –0.06), RAQoL −2.03 (−3.39, –0.69), SF-36 PCS +3.83 (1.49, 6.17) and 6MWT +56 (38, 75) m). One mild, transient AE occurred in the intervention group.ConclusionLongstanding, supervised exercise therapy was more effective than usual care in people with RA and severe functional limitations.Trial registration numberNetherlands Trial Register (NL8235), included in the International Clinical Trial Registry Platform (https://trialsearch.who.int/Trial2.aspx?TrialID=NL8235).