ObjectivesTo investigate whether and to what extent, return to work (RTW) expectancy and workability mediate the effect of two vocational interventions on reducing sickness absence in workers on sick leave from a musculoskeletal condition.MethodsThis is a preplanned mediation analysis of a three-arm parallel randomised controlled trial which included 514 employed working adults with musculoskeletal conditions on sick leave for at least 50% of their contracted work hours for ≥7 weeks. Participants were randomly allocated (1:1:1) to one of three treatment arms; usual case management (UC) (n=174), UC plus motivational interviewing (MI) (n=170) and UC plus a stratified vocational advice intervention (SVAI) (n=170). The primary outcome was the number of sickness absence days over 6 months from randomisation. Hypothesised mediators included RTW expectancy and workability assessed 12 weeks after randomisation.ResultsThe mediated effect of the MI arm compared with UC on sickness absence days through RTW expectancy was −4.98 days (−8.89 to −1.04), and workability was −3.17 days (−8.55 to 2.32). The mediated effect of the SVAI arm compared with UC on sickness absence days through RTW expectancy was −4.39 days (−7.60 to −1.47), and workability was −3.21 days (−7.90 to 1.50). The mediated effects for workability were not statistically significant.ConclusionsOur study provides new evidence for the mechanisms of vocational interventions to reduce sickness absence related to sick leave due to musculoskeletal conditions. Changing an individual’s expectation that RTW is likely may result in meaningful reductions in sickness absence days.Trial registration numberNCT03871712.