Behavioral activation (BA) has significant and long-lasting effects on depression in adults, and a growing body of evidence supports its efficacy with adolescents. It remains unclear, however, whether BA reduces systematically-observed depressive behavior in addition to subjective report of depressive symptoms. In this single-case study, three adolescents at a junior high school participated due to self-reported depression and directly-observed social withdrawal. They received 8 weeks of individual therapy at school focusing on the two core techniques of BA: activity scheduling and self-monitoring. To evaluate progress, participants completed the self-report RADS-2 once a week and had their socialization behavior observed during lunch three times a week. That is, trained coders used interval recording to track how much participants initiated/maintained eye contact or conversation with a peer/adult in the school's cafeteria. Comparing baseline to treatment termination, all three participants reported significant reductions in their RADS-2 Depression Total scales, and two participants exhibited meaningful increases in their socialization behavior. These results not only suggest that the efficacy of BA extends beyond traditional measures of depression, but also highlight the utility of systematic direct observation in identifying and monitoring specific depressive symptoms.