OBJECTIVES: The primary objective was to compare the effect of cognitive behavioural therapy for insomnia (CBT-I) to usual care on sleep efficiency, measured by polysomnography (PSG) immediately after the intervention at week 7. Secondary objectives included comparing the longer-term effect on sleep- and RA-related outcomes at week 26. METHODS: In a randomized controlled trial using a parallel group design, the experimental intervention was 6 weeks' nurse-led group-based CBT-I; the comparator was usual care. Analyses were based on the intention-to-treat (ITT) principle; missing data were statistically modelled using repeated-measures linear mixed effects models adjusted for the level at baseline. RESULTS: The ITT population consisted of 62 patients (89% women), with an average age of 58 years and an average sleep efficiency of 83.1%. At primary end point, sleep efficiency was 88.7% in the CBT-I group, compared with 83.7% in the control group (difference: 5.03 [95% CI -0.37, 10.43]; P = 0.068) measured by PSG at week 7. Key secondary outcomes measured with PSG had not improved at week 26. However, for all the patient-reported key secondary sleep- and RA-related outcomes, there were statistically highly significant differences between CBT-I and usual care (P < 0.0001), e.g. insomnia (Insomnia Severity Index: -9.85 [95% CI -11.77, -7.92]) and the RA impact of disease (RAID: -1.36 [95% CI -1.92, -0.80]) at week 26. CONCLUSION: Nurse-led group-based CBT-I did not lead to an effect on sleep efficiency objectively measured with PSG. However, CBT-I showed improvement on all patient-reported key secondary sleep- and RA-related outcomes measured at week 26. TRIAL REGISTRATION: ClinicalTrials.gov, https://clinicaltrials.gov, NCT03766100.