DURING THE STUDY PERIOD, 178 VIOLENT INCIDENTS WERE RECORDED, WITH 6 HOMES REPORTING OVER 10 INCIDENTS AND 2 HOMES RECORDING NO INCIDENTS. A MINORITY (44 PERCENT) OF THE CHILDREN WERE INVOLVED IN THE INCIDENTS, WITH 5 PERCENT ACCOUNTING FOR 37 PERCENT OF THE INCIDENTS. ASSULTS ON STAFF WERE THE MOST COMMON TYPE OF VIOLENT INCIDENT. THE THREE HOMES WITH DISPROPORITIONATE NUMBERS OF VIOLENT INCIDENTS HAD HIGHER CONCENTRATIONS OF CHILDREN FROM DISTURBED FAMILIES AND LARGER PROPORTIONS OF 11 TO 15 YEAR OLDS. IN ALL OF THE HOMES WHERE VIOLENCE OCCURRED, THE CHILDREN INVOLVED TENDED TO HAVE EXPERIENCED VIOLENCE IN THEIR FAMILIES AND WERE LIKELY TO BE BETWEEN THE AGES OF 11 AND 15; PARTICIPANTS WERE AS LIKELY TO BE GIRLS AS BOYS. HOMES WHERE THE MOST VIOLENCE OCCURRED ALSO HAD A LOWER RESIDENT-STAFF RATIO THAN HOMES WITH FEW OR NO INCIDENTS. WAYS OF CONTROLLING THE PROPORTION OF RESIDENTS AGED 11 TO 15 IN THE HOMES SHOULD BE CONSIDERED; INCREASED FOSTER CARE MIGHT HELP. THE SALARIES AND NUMBERS OF STAFF IN COMMUNITY HOMES SHOULD ALSO BE INCREASED TO REDUCE TURNOVER AND IMPROVE THE QUALITY OF CARE PROVIDED. TABULAR DATA AND REFERENCES ARE PROVIDED. (AUTHOR ABSTRACT MODIFIED--RCB)