Objective To evaluate safety and efficacy of long-term denosumab 60 mg every 6 (Q6M) or 3 months (Q3M) in rheumatoid arthritis (RA) patients. Methods This 12-month, randomised, double-blind, placebo-controlled, multicentre phase 3 trial with an open-label extension period from 12 to 36 months (DESIRABLE) enrolled Japanese RA patients treated with placebo for 12 months then denosumab Q6M (P/Q6M) or denosumab Q3M (P/Q3M); denosumab Q6M for 36 months (Q6M/Q6M); or denosumab Q3M for 36 months (Q3M/Q3M). Efficacy was assessed by van der Heijde modified total Sharp (mTSS), bone erosion (ES), and joint space narrowing (JSN) scores. Results Long-term treatment better maintained mTSS and ES suppression in the P/Q3M and Q3M/Q3M versus P/Q6M and Q6M/Q6M groups; changes from baseline in total mTSS at 36 months were 2.8 (standard error 0.4), 1.7 (0.3), 3.0 (0.4), and 2.4 (0.3), respectively; corresponding changes in ES were 1.3 (0.2), 0.4 (0.2), 1.4 (0.2), and 1.1 (0.2). No JSN effect was observed. Bone mineral density consistently increased in all groups after denosumab initiation, regardless of concomitant glucocorticoid administration. Serum C-telopeptide of type I collagen decreased rapidly at 1-month post-denosumab administration (both in the initial 12- month [Q3M, Q6M groups] and long-term treatment [P/Q3M, P/Q6M groups] phases). Adverse event incidence leading to study drug discontinuation was similar across treatment groups. Conclusion Denosumab treatment maintained inhibition of progression of joint destruction up to 36 months. Based on effects on ES progression, higher dosing frequency at an earlier treatment stage may be needed to optimise treatment. Denosumab was generally well tolerated.