FORCED SEXUAL COMPLIANCE, PARTICULARLY FOR PHYSICALLY WEAKER INMATES, IS A MAJOR PROBLEM IN PRISON TODAY. LITTLE PUBLISHED INFORMATION IS AVAILABLE WHICH FOCUSES ON THE SITUATION, OFTEN BECAUSE OF VICTIMS' RELUCTANCE TO INFORM ON OTHER INMATES FOR FEAR OF REPRISALS. IN THIS 12-MONTH STUDY, OVER 12 INDIVIDUALS--7 BLACKS AND 5 CHICANOS--WERE IDENTIFIED OUT OF A TOTAL OF 1,100 INMATES AT A FEDERAL CORRECTIONAL INSTITUTION AND SEGREGATED FOR HAVING RAPED OTHER INMATES. THESE INMATES WERE EVALUATED IN TERMS OF 48 VARIABLES IN THEIR FILES, SUCH AS YEARS OF FORMAL EDUCATION AND PRIOR ARRESTS. SEVEN BLACKS AND FIVE CHICANO INMATES WERE SELECTED AT RANDOM FROM THE SAME PRISON POPULATION FOR CONTROL PURPOSES. THE TOTAL OF 24 SUBJECTS WERE DIVIDED INTO FOUR GROUPS: BLACK-RAPIST, BLACK-NONRAPIST, CHICANO-RAPIST, AND CHICANO-NONRAPIST. A TWO-WAY MULTIVARIATE ANALYSIS OF VARIANCE WAS PERFORMED ON THESE GROUPINGS, WITH THE RAPIST CONDITION DIFFERING SLIGHTLY WITH THAT OF THE NONRAPIST. THE OBTAINED VALUES FOR BLACK VS. CHICANO AND FOR INTERACTION WERE LESS THAN THE CRITICAL VALUE. BECAUSE THE NUMBER OF RAPISTS POSSIBLE FOR THE STUDY WAS SMALL, ANY INFERENCE MUST BE MADE WITH CAUTION. HOWEVER, RESULTS SUGGEST THAT IT MAY BE POSSIBLE TO DIFFERENTIATE POTENTIAL MALES WHO MIGHT SUBJECT OTHER INMATES TO RAPE FROM NONRAPISTS ON THE BASIS OF INFORMATION ROUTINELY RECORDED IN EVERY INMATE'S FILE. WITH A LARGER SAMPLE, A MULTIPLE REGRESSION EQUATION COULD BE CONSTRUCTED FOR PREDICTION PURPOSES. IN THIS STUDY, ALL RAPISTS WERE MEMBERS OF MINORITY GROUPS, ALL VICTIMS BUT TWO WERE CAUCASIAN, AND ALL VICTIMS WERE OF A DIFFERENT RACE THAN THE RAPIST. FOOTNOTES, REFERENCES, AND TABLES ARE PROVIDED IN THE STUDY. (LWM)