The CSC study found that the universal delivery of a school-based, online programme for the prevention of mental health and substance use disorders among adolescents resulted in improvements in mental health and substance use outcomes at 30-month follow-up. We aimed to compare the long-term effects of four interventions-Climate Schools Combined (CSC) mental health and substance use, Climate Schools Substance Use (CSSU) alone, Climate Schools Mental Health (CSMH) alone, and standard health education-on mental health and substance use outcomes among adolescents at 72-month follow-up into early adulthood. This long-term study followed up adolescents from a multicentre, cluster-randomised trial conducted across three states in Australia (New South Wales, Queensland, and Western Australia) enrolled between Sept 1, 2013, and Feb 28, 2014, for up to 72 months after baseline assessment. Adolescents (aged 18-20 years) from the original CSC study who accepted contact at 30-month follow-up and provided informed consent at 60-month follow-up were eligible. The interventions were delivered in school classrooms through an online delivery format and used a mixture of peer cartoon storyboards and classroom activities that were focused on alcohol, cannabis, anxiety, and depression. Participants took part in two web-based assessments at 60-month and 72-month follow-up. Primary outcomes were alcohol use, cannabis use, anxiety, and depression, measured by self-reported surveys and analysed by intention to treat (ie, in all students who were eligible at baseline). This trial is registered with the Australian New Zealand Clinical Trials Registry (ACTRN12613000723785), including the extended follow-up study. Of 6386 students enrolled from 71 schools, 1556 (24·4%) were randomly assigned to education as usual, 1739 (27·2%) to CSSU, 1594 (25·0%) to CSMH, and 1497 (23·4%) to CSC. 311 (22·2%) of 1401 participants in the control group, 394 (26·4%) of 1495 in the CSSU group, 477 (37·%) of 1289 in the CSMH group, and 400 (32·5%) of 1232 in the CSC group completed follow-up at 72 months. Adolescents in the CSC group reported slower year-by-year increases in weekly alcohol use (odds ratio 0·78 [95% CI 0·66-0·92]; p=0·0028) and heavy episodic drinking (0·69 [0·58-0·81]; p<0·0001) than did the control group. However, significant baseline differences between groups for drinking outcomes, and no difference in the predicted probability of weekly or heavy episodic drinking between groups were observed at 72 months. Sensitivity analyses increased uncertainty around estimates. No significant long-term differences were observed in relation to alcohol use disorder, cannabis use, cannabis use disorder, anxiety, or depression. No adverse events were reported during the trial. We found some evidence that a universal online programme for the prevention of anxiety, depression, and substance use delivered in early adolescence is effective in reducing the use and harmful use of alcohol into early adulthood. However, confidence in these findings is reduced due to baseline differences, and we did not see a difference in the predicted probability of drinking between groups at 72-month follow-up. These findings suggest that a universal prevention programme in adolescence is not sufficient to have lasting effects on mental health and substance use disorders in the long term. In addition to baseline differences, substantial attrition warrants caution in interpretation and the latter factor highlights the need for future long-term follow-up studies to invest in strategies to increase engagement. Australian National Health and Medical Research Council.