The study population included the last 50 escapees from the Louisiana Department of Corrections during 1982 and a comparison group of 50 inmates whose prison identification numbers were sequentially nearest those of the escapees. This approach was assumed to provide an adequately random selection, because the prison identification numbers were assigned chronologically. The inmates had all received a battery of tests and a classification interview during their first 10 days at the adult reception and diagnostic center. Discriminant analysis of the data suggested the possible usefulness of two risk assessment models. The first model included six variables: race, juvenile arrest history, employment history, IQ score, and two scales on the Minnesota Multiphasic Personality Inventory (MMPI). The second model had four variables: race, juvenile arrest history, employment history, and the K scale on the MMPI. A followup validation study tested the 2 models using data from a random sample of 50 Louisiana escapees during 1981 and 1983. Each subject was classified into either a high-risk or a low-risk group using each of the two models. The six-variable model was found to be inaccurate, whereas the four-variable model appeared to be a useful screening instrument for escape risk. The Louisiana Department of Corrections is planning to use and further evaluate this model. Data tables, an appendix presenting four models, and 25 references are included.