THE MEASURES WERE COMPARED USING A SAMPLE OF ALL ADULT MALE INMATES ELIGIBLE FOR PARTICIPATION FOR WORK RELEASE DURING THE LAST 6 MONTHS OF 1969, WITH DATA FOR THIS SAMPLE DERIVED FROM AN EXPERIMENTAL STUDY CONDUCTED IN 1974. THE FBI DATA WERE OBTAINED FROM RAP SHEETS FOR A FOLLOWUP PERIOD OF 39 MONTHS. RECIDIVISM DATA WERE OBTAINED FROM THE FDC FOR THE SAME FOLLOWUP PERIOD. INFORMATION WAS ALSO OBTAINED FROM INTERVIEWS WITH FORMER INMATES CONDUCTED ABOUT 2 YEARS AFTER THEIR RELEASE FROM PRISON. CONSIDERABLE DISCREPANCIES WERE NOTED AMONG RECIDIVISM MEASURES. VARIATION WAS GREATEST BETWEEN CATEGORIES AND LEAST WITHIN CATEGORIES. ABOUT 69.7 PERCENT OF THOSE REPORTED TO BE REINCARCERATED BY THE FBI WERE ALSO INCLUDED IN STATISTICS OF THE FDC. TREATING SELF-REPORTED ITEMS AS THE BASELINE, IT WAS AGAIN FOUND THAT THE GREATEST VARIATION WAS BETWEEN CATEGORIES. ALMOST 95 PERCENT OF THOSE INTERVIEWED WHO STATED THEY HAD BEEN REINCARCERATED WERE INCLUDED IN FBI STATISTICS (RECHARGED OR REARRESTED). PERCENTAGES OF THOSE RECIDIVATING ACCORDING TO FBI AND FDC MEASURES WERE ALMOST EQUAL. TWO OF THE MOST SIMILAR MEASURES, FBI REINCARCERATION AND FDC IMPRISONMENT (BOTH OF FELONS ONLY), IDENTICAL EXCEPT FOR DEPENDENT VARIABLES, ARE FITTED TO LOG-LINEAR MODELS. FINDINGS VARY GREATLY. IT IS CONCLUDED THAT RECIDIVISM STUDIES SHOULD BE INTERPRETED WITHIN A CONTEXT THAT REFLECTS CRITERIA FOR MEASURING RECIDIVISM, THE SAMPLE, AND THE LENGTH OF TIME FOR FOLLOWUP, SINCE ALL RECIDIVISM MEASURES ARE NOT NECESSARILY EQUALLY VALID OR RELIABLE AND THE USE OF DIFFERENT MEASURES CAN PRODUCE DISCREPANT FINDINGS. SUPPORTING DATA, DETAILS OF THE ANALYSIS, AND REFERENCES ARE INCLUDED. (DEP)