We aimed to assess the effectiveness of 'Paesaggi di Prevenzione', a school-based prevention program delivered by trained teachers and designed to tackle smoking, alcohol misuse, dietary risks, and physical inactivity in adolescence. We evaluated the program between 2010 and 2013 with a two-arm, parallel-group, multicentre cluster randomized controlled trial in which schools were the units of randomization. We collected data on health-compromising behaviours using self-reported measurements of behaviour frequency administered before and after program implementation. We used multivariable mixed-effects logistic regression models to estimate program effects on health-compromising behaviours. The analysis sample included 3410 middle school students and 1651 high school students. Among middle school students, mean age at baseline was 12 years (standard deviation [SD] 0.5 years), 51% were boys, and 41% had high socioeconomic status [SES] (defined as having at least one parent/guardian with university level education). In high school students, mean age at baseline was 14 years (SD 0.7 years), 56% were boys, and 31% had high SES. The program did not have effects on smoking, alcohol misuse, and physical activity. The program had iatrogenic effects in regard to some eating behaviours, resulting in (i) lower odds of fruit consumption among middle-school students (odds ratio [OR] 0.82; 95% confidence interval [CI] 0.68-0.99) and (ii) lower odds of having breakfast every day in high-school students at the post-intervention measurement (OR 0.76; 95%CI 0.58-0.99) but not at one-year follow-up (OR 0.94; 95%CI 0.69-1.28). Due to the possibility of unintended effects, we advise against disseminating 'Paesaggi di Prevenzione' in its present form. TRIAL REGISTRATION: ISRCTN00953701.