Two types of variables which might be differentially associated with the performances of prison inmates while on furloughs were studied. They were, 1) demographic data in the central files of inmates known to have been successful or unsuccessful on furlough, and 2) personality test data taken on groups of inmates who could be expected to be either successful or unsuccessful on furlough, given the demographic data in the central files which is found to discriminate between the two groups. Demographic data was gathered from the files of adult male felons housed in correctional institutions in the following states: Virginia, Pennsylvania, and New York. The selection of subjects was limited to those inmates who were eligible for furlough between 1974 and 1977, and who had been approved and left the institution on at least one occasion. Personality test data was obtained from the administration of the Minnesota Multiphasic Personality Inventory, the California Psychological Inventory, and the S-0 Rorschach Test to 70 minimum security adult male felons, potentially eligible for furlough in the Virginia Department of Corrections. The SPSS Subprogram Discriminant, a linear discriminant analysis, run on the total number of known furlough success and failure cases, produced a discriminant function containing the following variables in order of their contributions to the total equation: Length of Term, To Whom Furloughed, Relationship to Parole Eligibility, Months Served Prior to Furlough, Emotional Maturity, Number of Parole Violations, Number of Prior Furloughs, and Marital Status. These variables correctly classified the furlough success and failure cases 73% of the time. Another linear discriminant analysis run on only those cases from the Virginia Department of Corrections produced a discriminant function containing the following variables in order of their contributions to the total equation: To Whom Furloughed, Marital Status, Number of Prior Furloughs, Months to Discharge Date, Total Number of Convictions, and Emotional Maturity. These variables correctly classified the furlough success and failure cases 71% of the time. The following personality test scales and/or factors showed significant differences in means between the potential furlough success and failure groups: Panton's special MMPI scales - Habitual Criminal Scale (HC), Prison Adjustment Scale (Ap), and Escapism Scale (Es); the CPI Factor II Scales - Responsibility (Re), Socialization (So), Tolerance (To), Good Impression (Gi), and the Factor III Scale - Achievement via Conformity (Ac); five S-0 Rorschach factors - F-, M, H, P, and Fch. The generalized conclusions drawn from these results suggest that it is possible to develop a discriminative function which can correctly classify inmates into furlough success and failure groups at least 70% of the time, on the basis of demographic variables found in inmate central files. Secondly, personality test data suggest that there are measurable differences in personality characteristics between inmates likely to succeed on furlough and those most likely to fail.