Whole-school interventions are a promising approach to preventing bullying and aggression while promoting broader health. The main analyses from a trial of the INCLUSIVE whole-school intervention reported reductions in bullying victimisation but not aggression and improved mental well-being. Latent transition analysis can examine how interventions 'move' people between classes defined by multiple outcomes over time. We examined at baseline what classes best defined individuals' bullying, aggression and mental well-being and what effects did the intervention have on movement between classes over time? INCLUSIVE was a two-arm cluster-randomised trial with 20 high schools per arm, with 24-month and 36-month follow-ups. We estimated sequential latent class solutions on baseline data. We then estimated a latent transition model including baseline, 24-month and 36-month follow-up measurements. Our sample comprised 8179 students (4082 control, 4097 intervention arms). At baseline, classes were (1) bullying victims, (2) aggression perpetrators, (3) extreme perpetrators and (4) neither victims nor perpetrators. Control students who were extreme perpetrators were equally likely to stay in this class (27.0% probability) or move to aggression perpetrators (25.0% probability) at 24 months. In the intervention group, fewer extreme perpetrators students remained (5.4%), with more moving to aggression perpetrators (65.1%). More control than intervention extreme perpetrators moved to neither victims nor perpetrators (35.2% vs 17.8%). Between 24 and 36 months, more intervention students moved from aggression perpetrators to neither victims nor perpetrators than controls (30.1% vs 22.3%). Our findings suggest that the intervention had important effects in transitioning students to lower-risk classes.