Research involved a representative sample of one-quarter of the male inmates (about 2,500) released from Canadian Federal penitentiaries in 1970, 1971, and 1972. Data were analyzed using regression analysis and predictive attribute analysis. Results confirmed previous findings that the seriousness of the crime was not an important discriminating factor in parole decisions when a minimum term had been set. In keeping with the Parole Board's policy statements and statutory mandates, risk was the prime consideration in parole decisionmaking. Findings on offender characteristics related to parole recidivism suggest that crimes against persons were associated with lower recidivism rates than property crimes. The simple summation instrument was determined to be the best means of correcting the discrepancy in best-risk decisions for both general and violent recidivism categories. In the proposed guidelines, risk is the key factor in parole decisions. Based on risk scores, inmates would qualify for particular parole procedures, with high-risk offenders being paroled later than low-risk offenders and under a series of conditional releases before final parole release.