An overview of prior research focuses on the relationship between the inmate's social and mental attributes and actual behavior inside the institution. Two approaches to developing structured classification systems are examined: consensus-based and actuarial models. Two classification scales are explained, the decision tree and the additive scale. The document describes five major classification models used by Illinois, Florida, New York, the National Institute of Corrections, and the Federal Bureau of Prisons. Despite the relatively primitive state of the art in classification, the report concludes that several important trends have emerged. For example, States adopting objective-based models have experienced reductions in the proportion of inmates assigned to maximum security and associated increases in minimum and medium levels of security. No associated increases in disciplinary problems or escapes have occurred. Staff acceptance has generally been favorable, and little validation research has been completed. Implications of these findings for California's system are discussed. Tables and sample forms.