The research examined 93 variables grouped into seven categories: demographic, drugs and alcohol, psychological, present criminal offenses, criminal history, institutional performance, and information pertaining to the temporary absence. Inmates who failed as well as those who returned to the institution on time were included. The base failure rate among inmates granted unescorted temporary absences was only 4.3 percent. A prior fraud, escape and unlawfully at large offenses, time served, and prognosis emerged as predictor variables. The failure groups tended to serve a somewhat shorter absolute time than the success group. The model showed significant incremental validity when used to predict TA success in conjunction with current administrative criteria. When applied to a group of inmates denied release, it revealed that many would have been predicted successes in terms of their index scores. The predictive index represents an incremental selection tool and is best used as an adjunct to clinical judgment. Tables, graphs, and 11 footnotes.