Adolescents experiencing social anxiety often engage in safety behaviors―covert avoidance strategies for managing distress (e.g., avoiding eye contact)―that factor into the development and maintenance of their concerns. Prior work supports the psychometric properties of the Subtle Avoidance Frequency Examination (SAFE), a self-report survey of safety behaviors. Yet, we need complementary methods for assessing these behaviors within contexts where adolescents often experience concerns, namely, interactions with unfamiliar peers. Recent work indicates that, based on short, direct social interactions with adolescents, individuals posing as unfamiliar peers (i.e., peer confederates) and without assessment training can capably report about adolescent social anxiety. We built on prior work by testing whether we could gather valid SAFE reports from unfamiliar untrained observers (UUOs), who observed adolescents within archived recordings of these short social interactions. A mixed clinical/community sample of 105 adolescents self-reported on their functioning and participated in a series of social interaction tasks with peer confederates, who also provided social anxiety reports about the adolescent. Based on video recordings of these tasks, trained independent observers rated adolescents’ observed social skills, and an additional set of UUOs completed SAFE reports of these same adolescents. Unfamiliar untrained observers’ SAFE reports (a) related to adolescents’ SAFE self-reports, (b) distinguished adolescents on clinically elevated social anxiety concerns, (c) related to trained independent observers’ ratings of adolescent social skills within interactions with peer confederates, and (d) related to adolescents’ self-reported arousal within these same interactions. Our findings support use of unfamiliar observers’ perspectives to understand socially anxious adolescents’ interpersonal functioning.