Thin slice studies are studies that examine judgments based on brief exposure to expressive behaviours or still images. Only one previous study has examined the prediction of outcomes within a law enforcement context from thin slices of a police-public encounter, and it demonstrated that experienced officers outperformed less experienced officers in terms of the quality, appropriateness, and accuracy of their predictions . The present study extends this research by examining how a range of factors -including operational years of experience and training, familiarity with the encounter, confidence in the prediction, and thin slice length -impact prediction accuracy. Participants with varying levels of police experience and training were recruited. Participants viewed 16 randomly ordered videos (half of these were 10 seconds and half were 30 seconds in length) depicting a thin slice of a police-public encounter. After each video, the participant was asked to predict whether the subject would harm or attempt to harm the officer(s). My results demonstrated that higher levels of training, greater familiarity, and greater confidence in one's predictions was associated with greater odds of providing an accurate response; operational years of policing experience was not associated with this outcome. My results also demonstrated that most of these variables' relationships with prediction accuracy disappear when examining longer thin slices (i.e., 30-second videos), and have slightly larger effects when examining shorter thin slices (i.e., 10-second videos). Finally, specialized police training, years of experience, and familiarity were, in turn, found to predict greater confidence in one's predictions. These findings and their implications are discussed.