The study analyzed SARA ratings in 2,681 adult male offenders. Offenders were quite heterogeneous with respect to the presence of individual risk factors and to overall perceived risk. SARA ratings significantly discriminated between offenders with and without a history of spousal violence in one sample and between recidivistic and nonrecidivistic spousal assaulters in another. SARA ratings showed good convergent and discriminant validity with respect to other measures related to risk for general and violent criminality. The primary discretion-based rating on the SARA, the Summary Risk Rating, had good reliability and validity. The format of the SARA, which requires explicit consideration and coding of a fixed set of risk factors, may provide a sophisticated and flexible framework within which to exercise professional discretion. Alternatively, the SARA may facilitate the reliable coding of professional judgment, which then can be subjected to appropriate statistical analysis. Tables, notes, references