Background: This study evaluated the effectiveness of a comprehensive mental health literacy (MHL) educational programme known as “Sanita” for improving junior high school students’ knowledge of mental illness, attitudes towards people with mental health problems, and help-seeking behaviour. Methods: A randomised controlled trial with a parallel-group design was conducted. A total of 125 students (51 in the intervention group and 74 in the control group) received three 50-min classes and completed self-report questionnaires (Mental Illness and Disorder Understanding Scale, MIDUS; Reported and Intended Behaviour Scale, RIBS-J; and an original questionnaire investigating help-seeking behaviour) before and after the programme and three months later. Results: Regarding MIDUS, the post-test and 3-month follow-up test results showed a significant main effect of time-by-group interactions in a linear mixed model. Regarding RIBS-J, the post-test results showed a significant main effect of time-by-group interactions; however, the 3-month follow-up test showed no significant effect. No significant effects of time-by-group interactions were seen in the post-test and 3-month follow-up test results for help-seeking behaviour in a logistic regression-mixed model. Conclusions: The Sanita MHL educational programme was longitudinally effective at improving junior high school students’ knowledge of mental illness, although improvements in attitudes and help-seeking behaviour were insufficient.