THE INITIAL HYPOTHESIS OF THIS STUDY WAS THAT PAROLEES VIEWED AS 'DANGEROUS' WOULD BE MOST LIKELY TO BE RETURNED TO PRISON. 'DANGEROUS' WAS TYPIFIED BY SUCH COMMITMENT OFFENSES AS WILLFUL HOMICIDE, AGGRAVATED ASSAULT, AND FORCIBLE RAPE. IT WAS FOUND THAT 27.1 PERCENT OF ALL PAROLEES WERE RETURNED TO PRISON THE FIRST YEAR, BUT THAT THE VIOLATION RATE WAS 19.6 PERCENT FOR THOSE CONVICTED OFFENSES AGAINST PERSONS COMPARED TO 31.5 PERCENT FOR THOSE CONVICTED OF PROPERTY OFFENSES. MOREOVER, A LARGER PROPORTION OF THE PERSON OFFENDERS WERE CONVICTED OF TECHNICAL PAROLE VIOLATIONS. THE PERCENTAGE RETURNED TO PRISON FOR TECHNICAL VIOLATIONS, AS OPPOSED TO COMMITMENT OF ADDITIONAL CRIMES, RANGED FROM 88.4 PERCENT FOR THOSE CONVICTED OF WILLFUL HOMICIDE AND 84 PERCENT FOR THOSE CONVICTED OF NEGLIGENT MANSLAUGHTER TO 69 PERCENT FOR BURGLARY OFFENDERS AND 66.4 FOR AUTO THEFT OFFENDERS. FURTHER ANALYSIS OF THE DATA INDICATES THAT PROPERTY OFFENDERS WERE ALLOWED TO REMAIN IN THE COMMUNITY EVEN WHEN SHOWING 'SIGNS OF DETERIORATION IN PAROLE ADJUSTMENT.' MEANWHILE, 'DANGEROUS' OFFENDERS WERE HELD TO MUCH STRICTER STANDARDS OF BEHAVIOR. A HIERARCHY OF 'DANGEROUSNESS' BASED ON PAROLE VIOLATION PATTERNS IS DISCUSSED. IT IS SUGGESTED THAT A STUDY SAMPLE BE CONSTRUCTED WHICH APPLIES BOTH MORE STRINGENT AND LESS STRINGENT CRITERIA FOR ALL COMMITMENT OFFENSE GROUPS AND THE RESULTS EVALUATED. IT IS HYPOTHESIZED THAT PROPERTY OFFENDERS MAY PROVE TO BE MORE 'DANGEROUS,' AS MEASURED BY RECIDIVISM. FOOTNOTES CONTAIN REFERENCES. GRAPHS PRESENTING THE PERCENTAGE OF ALL PAROLE VIOLATIONS BY OFFENSE CATEGORY ARE APPENDED. (GLR)