The amount of violent media that is consumed on a daily basis by the average American and the empirically proven effects associated with such regular consumption have led scholars to consider violent media a public health threat, the risks of which, the public may not even fully appreciate (Huesmann, Dubow, & Yang, 2013). Previous research in the field of public health communication has found that different forms of evidence in public health risk messages are more or less effective in changing behavior depending on individual recipient characteristics (de Wit, Das & Vet, 2008; Reinard, 1988; Slater & Rouner, 1996). The present research investigated the effectiveness of different forms of evidence (narrative or statistical) in decreasing violent media consumption by increasing an individual's risk perceptions and negative attitude associated with violent media. In accordance with the Theory of Planned Behavior (Ajzen, 1991), it was hypothesized that these risk perceptions and attitudes would predict intentions toward violent media consumption, as well as subsequent consumption. The study was conducted via MTurk with a sample of one hundred and fifty participants (53% Male). Results showed that an individuals' violent media consumption predicted their attitude toward violent media (p = .035), and that their risk perception and attitude toward violent media significantly predicted their intentions to decrease violent media consumption (ps < .05). Though no significant difference was found between the effect of narrative and statistical evidence on general violent media consumption, exploratory analyses of effects on specific forms of media showed that narrative evidence resulted in a significant decrease in violent video game consumption (p = .042). Additionally, age predicted risk perception, the older the participant the less risk they perceived in violent media consumption (p = .010). Future research should investigate the effect of including all elements of the Theory of Planned Behavior on the ability of different evidence types to change behavior, and perhaps extend the time frame within which change is measured in order to maximize the ability to observe any true change in behavior.