PURPOSE: Interventions to improve school ethos can reduce substance use but "upstream" causal pathways relating to implementation and school-level changes are uncertain. We use qualitative and quantitative data from a pilot trial to build hypotheses regarding these. METHODS: The Healthy School Ethos intervention involved two schools being provided with facilitation, training, and funding to plan and implement actions (some mandatory and some locally determined) to improve school ethos over one year. The evaluation involved a pilot-trial with two intervention and two comparison schools; semi-structured interviews with facilitators, staff, and students; and baseline and follow-up surveys with students aged 11 to 12 years. RESULTS: Student accounts linked participation in planning or delivering intervention activities with improved self-regard and relationships with staff and other students. Some activities such as re-writing school rules involved broad participation. Students in receipt of actions such as peer-mediation or motivational sessions reported benefits such as improved safety and relationships. Some student accounts linked improved self-regard and relationships with increased engagement and aspirations, and reduced substance use. At 9-month follow-up, students in intervention schools reported less hurting and teasing of others and feeling unsafe at school. Other outcomes suggested intervention benefits but were not significant. CONCLUSIONS: School-ethos interventions may reduce substance use through upstream pathways involving the aforementioned factors. Future phase-III trials should quantitatively model the extent to which these mediate intervention effects.