Objective Successful prevention of eating disorders represents an important goal due to damaging long-term impacts on health and well-being, modest treatment outcomes, and low treatment seeking among individuals at risk. Mindfulness-based approaches have received early support in the treatment of eating disorders, but have not been evaluated as a prevention strategy. This study aimed to assess the feasibility, acceptability, and efficacy of a novel mindfulness-based intervention for reducing the risk of eating disorders among adolescent females, under both optimal (trained facilitator) and task-shifted (non-expert facilitator) conditions. Method A school-based cluster randomized controlled trial was conducted in which 19 classes of adolescent girls (N-=-347) were allocated to a three-session mindfulness-based intervention, dissonance-based intervention, or classes as usual control. A subset of classes (N = 156) receiving expert facilitation were analyzed separately as a proxy for delivery under optimal conditions. Results Task-shifted facilitation showed no significant intervention effects across outcomes. Under optimal facilitation, students receiving mindfulness demonstrated significant reductions in weight and shape concern, dietary restraint, thin-ideal internalization, eating disorder symptoms, and psychosocial impairment relative to control by 6-month follow-up. Students receiving dissonance showed significant reductions in socio-cultural pressures. There were no statistically significant differences between the two interventions. Moderate intervention acceptability was reported by both students and teaching staff. Discussion Findings show promise for the application of mindfulness in the prevention of eating disorders; however, further work is required to increase both impact and acceptability, and to enable successful outcomes when delivered by less expert providers.