DURING THE PERIOD STUDIED, THE NUMBER OF ASSAULTS NEITHER ROSE NOR DECLINED SIGNIFICANTLY. THE AVERAGE NUMBER OF ASSAULTS EACH MONTH WAS LESS THAN 1 PER 100 INMATES. THERE WAS AN INVERSE RELATIONSHIP BETWEEN THE NUMBER OF GUARDS AND THE INCIDENCE OF ATTACKS. FLUCTUATIONS IN THE PRISONER-PER-GUARD RATIO EXPLAINED 49 PERCENT OF THE VARIATION IN ASSAULT RATES FROM INSTITUTION TO INSTITUTION. THE CHARACTER OF THE INMATE POPULATION WAS ANOTHER DETERMINANT OF ASSAULT RATES. THE INSTITUTION WITH THE HIGHEST ASSAULT RATE HOUSES YOUNG FIRST OFFENDERS, WHO TEND TO BE LESS ADJUSTED TO INSTITUTIONAL LIVING. THE INSTITUTION WITH THE SECOND HIGHEST ASSAULT RATE IS THE STATE PENITENTIARY, WHICH HOUSES LONG-TERM PRISONERS REQUIRING MAXIMUM SECURITY. IT APPEARS THAT INMATE CLASSIFICATION PROCEDURES PREDETERMINE THAT INMATES PRONE TO BEHAVIORAL PROBLEMS WILL BE PLACED TOGETHER WITHIN THE SAME INSTITUTIONS. THE NEED FOR BETTER DATA ON WHICH TO BASE STUDIES OF INMATE VIOLENCE IS NOTED. IT IS RECOMMENDED THAT INSTITUTIONS KEEP MONTHLY LOGS OF INFORMATION PERTAINING TO SUCH VIOLENCE (DATES OF INSPECTIONS, ACTS OF VIOLENCE, GRIEVANCES, DISCIPLINARY ACTIONS, ETC.). SUPPORTING DATA ARE INCLUDED. (LKM)