Research shows that academic performance and adolescent mental health are associated, but it remains unclear whether, and in what ways, they reciprocally influence each other over time. This study aimed to examine reciprocal longitudinal associations between academic performance and mental health in adolescence, focusing on internalizing and externalizing problems; and to test whether associations vary across school years or by gender. Self-reported data were obtained from 3473 pupils (48.3% boys) followed annually from grade 7 to grade 9 in Swedish lower secondary school (ages 13–16), with mental health reassessed again in grade 10, the first year of upper secondary school. Academic performance was measured by end-of-year subject grades, and internalizing and externalizing problems were measured with the Strengths and Difficulties Questionnaire (SDQ). We estimated conventional cross-lagged panel models (CLPMs) and random-intercept CLPMs (RI-CLPMs) and used multigroup models to examine gender differences. CLPMs and RI-CLPMs yielded similar substantive conclusions. Externalizing problems showed clear bidirectional within-person links with grades: higher-than-usual externalizing predicted lower subsequent grades, and lower-than-usual grades predicted subsequent increases in externalizing. In contrast, internalizing symptoms showed no within- or between-person associations with grades. Associations did not differ systematically across school years or by gender. We conclude that, across lower secondary school, links between academic performance and mental health are reciprocal yet domain-specific. Behavior appears to be more tightly coupled with grades than mood, indicating that not all facets of mental health, nor all aspects of performance, are equally responsive to one another.