Most clinicians-in-training learn to summarize suicide risk in a categorical probability judgment expressed as low, moderate, or high, often with gradations like low-moderate. But what do we really mean when we say a patient is at “low” “moderate” or “high” risk? Risk compared to whom? Compared to when? In what setting? These labels are devoid of context, lack predictive validity, and provide little help when developing plans and responses to prevent suicide. In this Suicide Prevention Resource Center Director's Corner, Anthony Pisani, PhD explains moving away from a prediction model to a prevention model.