IN USING RECIDIVISM AS AN OUTCOME MEASUREMENT FOR REHABILITATION PROGRAMS, THE GREATEST NEED IS FOR MORE UNIFORM DEFINITIONS AND MEASURES OF RECIDIVISM, AN ISSUE ADDRESSED BY SOME OF THE FOLLOWING RECOMMENDATIONS: (1) FUTURE RECIDIVISM STUDIES SHOULD USE FEDERAL BUREAU OF INVESTIGATION REARREST AND REINCARCERATION INFORMATION, BECAUSE THESE DATA ARE READILY AVAILABLE AND WOULD ALLOW GREATER COMPARABILITY NATIONWIDE IN FUTURE STUDIES; (2) AN APPROPRIATE GROUP OF EXPERTS SHOULD BE CONVENED TO DETERMINE WHAT KINDS OF OFFENSES TO INCLUDE WHEN MEASURING RECIDIVISM, OR THERE WILL BE CONTINUED CONFUSION IN COMPARING RESULTS; (3) THE USE OF CONTINUOUS MEASURES OF RECIDIVISM SHOULD BE MORE FULLY EXPLORED, BECAUSE THIS WOULD ALLOW DETERMINATION OF THE RELATIVE SUCCESS OR FAILURE OF PARTICULAR INDIVIDUALS, WHILE TAKING ADVANTAGE OF INFORMATION IGNORED BY DICHOTOMOUS MEASURES; (4) FOLLOWUP PERIODS IN RECIDIVISM STUDIES SHOULD RANGE FROM A MINIMUM OF 3 YEARS TO A MAXIMUM OF 5 YEARS TO ASSURE THAT TREATMENT EFFECTS ARE NOT EXTINGUISHED WITHIN A FEW YEARS; (5) THERE SHOULD BE A CONTINUED RELIANCE ON OFFICIAL MEASURES OF RECIDIVISM AND SELF-REPORT MEASURES SHOULD BE USED WHEN POSSIBLE; (6) GREATER ATTENTION SHOULD BE FOCUSED ON THE RELIABILITY AND VALIDITY OF RECIDIVISM MEASURES AS A REFLECTION OF THE DIFFICULTIES OF MEASURING CRIME AND THE LACK OF ADEQUATE CRITERION VARIABLES FOR ESTABLISHING PREDICTIVE VALIDITY; AND (7) AT A MINIMUM, STUDIES OF RECIDIVISM SHOULD ONLY BE COMPARED WHEN THE SAMPLES, THE LENGTH OF FOLLOWUP, THE QUALITY OF THE RESEARCH DESIGN, AND METHODS OF RECIDIVISM MEASUREMENT ARE SIMILAR. REFERENCES ARE PROVIDED. (AUTHOR ABSTRACT MODIFIED--RCB)