In this article, we introduce multimodal social relations analysis as a powerful tool for studying personality pathology that tackles several important limitations of existing research. By implementing a design in which groups of participants provide repeated ratings as they interact, researchers can gather data on individuals’ mutual perceptions, affective experiences, and interpersonal behaviors in naturalistic social contexts. We demonstrate how the social relations model can be used to analyze and make conceptual sense of these complex, dyadic data and showcase how this may be used to address not only the experiences and behaviors of individuals diagnosed with a personality disorder but also the reactions these individuals evoke in others. We provide recommendations as to what settings and measures might be best suited when designing a study that applies multimodal social relations analysis, and we discuss practical and theoretical implications as well as possible extensions to this method.