Video data analysis (VDA) represents an important methodological framework for contemporary research approaches to the myriad of footage available from cameras, devices, and phones. Footage from police body-worn cameras (BWCs) is anticipated to be a widely available platform for social science researchers to scrutinize the interactions between police and citizens. We examine issues of validity and reliability as related to BWCs in the context of VDA, based on an assessment of the quality of audio and video obtained from that platform. Second, we compare the coding of BWC footage obtained from a sample of police-citizen encounters to coding of the same events by on-scene coders using an instrument adapted from in-person systematic social observations (SSOs). Findings show that there are substantial and systematic audio and video gaps present in BWC footage as a source of data for social science investigation that likely impact the reliability of measures. Despite these problems, BWC data have substantial capacity for judging sequential developments, causal ordering, and the duration of events. Thus, the technology should open theoretical frames that are too cumbersome for in-person observation. Theoretical development with VDA in mind is suggested as an important pathway for future researchers in terms of framing data collection from BWCs and also suggesting areas where triangulation is essential.