Adolescent depression constitutes an escalating global health issue, defined by complex interactions among biological, psychological, and social determinants. This narrative review synthesizes contemporary evidence on the effects of physical exercise as a non-pharmacological intervention for depressive symptoms in adolescents. Emerging empirical findings from randomized controlled trials suggest that exercise may provide mental health benefits across diverse adolescent populations, including individuals with depression and related comorbid conditions such as attention-deficit/hyperactivity disorder, obesity, and cardiometabolic disorders. Exercise interventions, particularly those involving moderate-to-vigorous intensity and multimodal approaches, have been associated with reductions in depressive symptoms, anxiety, and stress, while also improving quality of life, resilience, and cognitive performance. Mechanistically, these effects are thought to be partially mediated by exercise-induced neurobiological changes, including modulation of brain-derived neurotrophic factor, stress-related biomarkers such as cortisol, and myokine signaling pathways, which may support neuroplasticity and emotional regulation. However, direct evidence for epigenetic mechanisms in adolescent populations remains limited and should be interpreted with caution. Furthermore, digital mental health interventions integrated with physical activity may offer scalable preventive strategies within school-based and resource-limited settings. Nevertheless, substantial heterogeneity in outcomes persists, influenced by factors such as exercise intensity, adherence, baseline symptom severity, and individual genetic and environmental differences. While some studies report minimal or delayed effects, the overall evidence supports exercise as a promising complementary or preventive approach for adolescent depression. This review is a narrative synthesis, and future research should prioritize standardized methodologies, longitudinal designs, and mechanistic investigations to strengthen causal inference and translational applicability.