THE DEVICE IN QUESTION IS THE SALIENT FACTOR SCORE--AN INDEX OF THE LIKELIHOOD THAT A PAROLE APPLICANT WILL COMMIT A NEW OFFENSE OR VIOLATE PAROLE. THE SCORE TAKES INTO ACCOUNT SUCH FACTORS AS THE OFFENDER'S PAST INVOLVEMENT WITH THE CRIMINAL JUSTICE SYSTEM, HISTORY OF DRUG ABUSE, EDUCATION AND EMPLOYMENT HISTORY, AND RELEASE PLANS. A SAMPLE OF 1,138 CASES, ONE-THIRD OF ALL FEDERAL INMATES SERVING MAXIMUM SENTENCES OF MORE THAN 1 YEAR WHO WERE RELEASED DURING THE LAST 6 MONTHS OF 1971, WAS FOLLOWED FOR A 3-YEAR PERIOD. THE SAMPLE INCLUDED ALL MAJOR FORMS OF RELEASE (PAROLE, MANDATORY RELEASE, EXPIRATION OF SENTENCE). FEDERAL BUREAU OF INVESTIGATION RECORDS WERE USED TO DETERMINE SUCCESS/FAILURE OUTCOMES IN TERMS OF THREE CRITERION MEASURES: NEW ARREST, NEW CONVICTION, AND NEW COMMITMENT. THE RESULTS WERE COMPARED WITH RESULTS FROM VALIDATION TESTS BASED ON 2-YEAR FOLLOWUPS OF SAMPLES OF FEDERAL INMATES RELEASED IN 1970 AND 1972. TOGETHER, THE STUDIES INDICATE THAT THE SALIENT FACTOR SCORE IS REMARKABLY STABLE IN ITS ABILITY TO ASSIGN CASES TO DISTINCT 'RISK' GROUPS RELATIVE TO EACH OF THE CRITERION MEASURES EXAMINED. THE CHOICE OF CRITERION MEASURE DOES NOT APPEAR TO HAVE MUCH EFFECT IN THE ASSESSMENT OF THE DEVICE'S PREDICTIVE POWER. HOWEVER, THERE ARE INDICATIONS THAT THIS CHOICE MAY HELP TO EXPLAIN CONFLICTING REPORTS AS TO WHETHER RELEASEES WHO SUCCEED IN THE FIRST YEAR OR 2 AFTER RELEASE BECOME SUBSTANTIALLY LESS LIKELY TO FAIL IN SUBSEQUENT YEARS. SUPPORTING DATA, A COPY OF THE SALIENT FACTOR SCORING DEVICE, AND A COPY OF THE INSTRUMENT USED BY THE COMMISSION TO CALCULATE OFFENSE SERIOUSNESS ARE INCLUDED. (LKM)