Background. Autonomy-supportive implementation intention exercises have been shown to facilitate goal-directed behaviour (Koestner et al., 2006). The current study explored whether eating self-efficacy moderated the impact of autonomy-framed versus control-framed implementation intentions to reduce high-calorie snack intake. Methods. The study employed a randomized prospective design, involving two waves of data collection conducted in 2016. At Time 1, UK participants (N = 300) completed an online questionnaire which asked them to report their snacking behaviour over the previous 7 days. Participants were subsequently asked to form either an autonomy-framed implementation intention or a control-framed implementation intention. Seven days later, participants reported their consumption of high-calorie snacks and completed a measure of eating self-efficacy. Results. Hierarchical multiple regression analysis revealed that eating self-efficacy moderated the effects of implementation intention framing. Autonomy-framed implementation intentions had a greater impact on the avoidance of snacking for high eating self-efficacy participants than did control-framed implementation intentions. In contrast, for low eating self-efficacy participants, control-framed implementation intentions had more impact than did autonomy-framed implementation intentions. Conclusions. The results suggest that if implementation intentions to promote healthy diet are to be effective, the role of eating self-efficacy should be considered, and the design of interventions adapted accordingly.