Previous research on adolescent romantic relationships has been largely based on self-reports and interview data; as a result, relatively little is known about the interpersonal-behavioral dynamics of adolescent couples. In an attempt to address this gap in the previous literature on young couples, the present study used observational methods to differentiate between healthy and dysfunctional adolescent romantic relationships. Two groups of adolescent couples were recruited to participate in this study: (1) a high-risk group (n=18 couples) in which one or both partners had a substance use disorder (SUD) and (2) a low-risk group (n=12 couples) in which neither partner had a history of psychopathology. Self-report and observational data on couples' relationships were collected from both groups. Couples' observed conflict interactions were coded using the structural analysis of social behavior [Florsheim, P., & Benjamin, L. S. (2001). The structural analysis of social behavior observational coding scheme. In P. K. Kerig, & M. Lindahl (Eds.), "Family observational coding schemes: Resources for systemic research" (pp. 127-150). Hillsdale, NJ: Lawrence Erlbaum Associates]. Findings indicated that, compared to couples with no psychopathology, couples in the SUD group engaged in significantly more hostile and less warm behavior, as well as more complex communication involving a mix of hostility and warmth. Self-reported relational quality did not differentiate the two groups, highlighting the unique contributions of observational data for understanding the clinically relevant dynamics of adolescent romantic relationships. (Contains 1 figure and 2 tables.)