The majority of studies using observational coding systems for family interaction data derive scales describing family members’ behaviors based on rational/theoretical approaches. This study explored an empirical approach to identifying the component structure of parent–child observational data that incorporated the affective context of the interaction. Dyads of 155 typically developing 8-year-olds and their mothers completed questionnaires and two interaction tasks, one each designed to illicit positive and negative interactions. Behaviors were coded based on a modified version of the Family Interaction Macro-coding System. Multiple factor analysis identified four-component solutions for the maternal and child data. For both, two of the components included negative behaviors, one positive behavior, and one communicative behavior. Evidence for the validity of the maternal and child components was demonstrated by associations with child depression and anxiety symptoms and behavioral problems. Preliminary evidence supports an empirical approach to identify context-specific components in parent–child observational data.