The purpose of this study was to observe how within-person variation in motivation, self-efficacy and academic performance scores are correlated or can be predicted. The reciprocal association between these variables was analysed through a four-wave longitudinal study and a within-person analytical approach (random intercept cross-lagged panel model, RI-CLPM). A total of 932 adolescents, 417 boys and 515 girls (M[subscript age] = 14.34 years, SD = 0.90) enrolled in the first year of high school (9th grade) participated in the study. Questionnaires on motivation and self-efficacy were administered once a year for four years. At the end of each school year, teachers' evaluations of students' academic performance, expressed in school grades, were collected. The cross-lagged paths of the RI-CLPM suggested that higher-than-expected self-reported levels of self-efficacy were associated with above-average levels of motivation one year later and vice versa. In addition, adolescents who reported higher-than-expected levels of self-efficacy showed above-average levels of academic performance one year later and vice versa, and only adolescents who reported higher-than-expected levels of academic performance had above-average levels of motivation, but not vice versa. However, a mediation analysis revealed an indirect role of motivation in academic performance through self-efficacy. Implementing projects that strengthen motivation and self-efficacy can thus reduce the risk of poor academic performance and leaving school. Preventing young people from dropping out of school can help promote lifelong learning opportunities for all.