A variety of instruments have been published over recent years that improve clinicians’ ability to forecast the likelihood that an individual will behave violently. Increasingly, these instruments are being applied in response to laws that require specialized risk assessments. In this article, we present a framework that goes beyond the "clinical" and "actuarial" dichotomy to describe a continuum of structured approaches to risk assessment. Despite differences among them, there is little evidence that one validated instrument predicts violence better than another. We believe that these group-based instruments are useful for assessing an individual’s risk, and that an instrument should be chosen based on an evaluation’s purpose (i.e., risk assessment vs. risk reduction). The time is ripe to shift attention from predicting violence to understanding its causes and preventing its (re)occurrence.