Background: Existing evidence indicates that social support may enhance recipients' self-efficacy (enabling hypothesis) or that self-efficacy facilitates support receipt (cultivation hypothesis). However, less is known about the time-lagged support-self-efficacy relationship in couples. Our aim was to disentangle reciprocal interrelations among stable and time-varying components of support provision and self-efficacy in couples over time. Methods: We conducted secondary analyses of a published randomised controlled trial with six assessments, spanning 1 year and N = 338 heterosexual couples (age range: 18-80 years). Women's and men's reports on physical activity-specific provided support and physical activity-specific self-efficacy were analysed. Results: Based on the actor-partner interdependence bs_bs_banner model, we compared nested random intercepts cross-lagged panel models. The final model revealed no gender effects. Stable levels of both partners' support provision and self-efficacy were positively associated. At the time-varying level, one partner's self-efficacy predicted the other partner's support provision later on. No lagged-association emerged for the opposite predictive direction. Conclusions: Partners' stable shares of provided support and self-efficacy were interrelated, whereas higher time-varying self-efficacy of one partner seemed to activate support provision from the other partner, confirming the cultivation hypothesis but not the enabling hypothesis.