Background <p>While dance gains recognition as a health resource for older adults, empirical evidence remains inconclusive. Personal factors likely to influence participation in structured dance programs have been overlooked, potentially contributing to the conflicting results. Like exercise self-efficacy, dance self-efficacy may affect engagement with group dancing considering the physical, cognitive, and social challenges involved. This study examines the associations between dance self-efficacy, program participation and satisfaction, and health outcomes.</p> Methods <p>Participants of the DAnCE cluster-RCT in Sydney, Australia, (<i>n</i> = 217) received folk or ballroom dancing, twice weekly for 12 months. Quantitative measures of dance self-efficacy, program factors (attendance, satisfaction, and difficulty ratings), the Short Physiological Performance Battery (SPPB), Rey Auditory Visual Learning Tasks, and Trail Making Tests A&amp;B were completed. ANCOVA investigated associations between dance self-efficacy and program factors. Linear regression models, adjusted for co-variates, examined dance self-efficacy as a moderator and mediator of the effect of attendance on physical (SPPB) and cognitive (composite) health outcomes.</p> Results <p>Participants mean (SD) age was 79 years (7.19). High dance program attendance (≥ 56/85 sessions) was associated with improved dance self-efficacy, while low attenders (≤ 25/85 sessions) decreased in dance self-efficacy (F = 10.85, η² = 0.11, <i>p</i> &lt; .01). Both high attendance and increased dance self-efficacy were associated with greater program satisfaction (<i>p</i> &lt; .01) and lower perceived difficulty (<i>p</i> &lt; .01). Baseline dance self-efficacy moderated the effect of program attendance on physical function (SPPB; <i>p</i> = .03 for interaction) such that only participants with low dance self-efficacy at baseline realised physical functional gains. Change in dance self-efficacy fully mediated effect of attendance on physical function outcomes (b = 0.20, BCa 95%CI: 0.04, 0.41), but not cognitive outcomes.</p> Conclusions <p>Dance self-efficacy appears to influence both dance program participation and physical function outcomes, suggesting it may be a key mechanism driving health benefits. Incorporating a dance self-efficacy tool in older adult dance research and practice may improve initial assessments of dance program participant needs, guide program design and evaluation, and lead to more consistent health outcomes.</p> Trial registration <p>The Australian New Zealand Clinical Trials Registry ACTRN12612000889853 (21.08.2012).</p>