The relationship between parent and child is one that has been studied intensively for years. Much of the previous research in this field has quantified the parent-child relationship through self-report measures, with a subsample coding behavior from videotape and averaging individual scores across the entire parent-child interaction. Using a dynamic systems approach, we attempted to gain a deeper understanding of the parent-child relationship by quantifying the relationship in terms of dyadic patterns using the software Gridware. We then used these dyadic patterns to predict internalizing and externalizing behaviors in eight-year-old twin children. Dyadic relationship patterns predicted externalizing behaviors such as aggression and conduct disorder (i.e., frequency and stability within negative attractor states, and infrequency and low stability in positive attractor states), but not internalizing behaviors. Findings provide a method for capturing variance in parent-child interactions that is important for children's externalizing behaviors. Future studies should utilize these patterns in understanding risk and resilience family processes for children's mental health and well being.