THE 48 INCIDENTS PERFORMED BY 39 WOMEN ARE ALSO CONSIDERED, TO DETERMINE WHETHER THERE WAS ANY CONSISTENCY IN THE TIME AND PLACE OF OCCURRENCE, OR ANY PATTERN IN THE PRECEDING AND FOLLOWING EVENTS. FACTORS TAKEN INTO CONSIDERATION INCLUDED AGE, MARITAL STATUS, RELIGION, PREVIOUS INSTITUTIONALIZATION, PREVIOUS SELF-INJURY, SENTENCE, OFFENSE, RESULTS OF A PERSONALITY INVENTORY MEASURING HOSTILITY AND DIRECTION OF HOSTILITY, LOCATION OF SELF-INJURY INCIDENT, NATURE OF INCIDENT, MENSTRUAL CYCLE, PRECIPITATING EVENTS, FOLLOWING EVENTS, NUMBERS OF WOMEN AND STAFF ON DUTY, AND DATES OF SELF-INJURY INCIDENTS. THE INCIDENTS OCCURRED MOSTLY IN A CELL, AND MOST CONSISTED OF CUTTING THE ARM, GENERALLY WITH A PIECE OF GLASS. THE INCIDENTS TENDED TO OCCUR IN RUNS OR 'EPIDEMICS.' THE MOST SIGNIFICANT FINDINGS ARE THAT THE SELF-MUTILATORS TEND TO BE YOUNGER THAN THE PRISON NORM, HAVE A HIGH DEGREE OF PREVIOUS INSTITUTIONAL AND PSYCHIATRIC EXPERIENCE, HAVE COMMITTED MORE VIOLENT OFFENSES THAN A CONTROL GROUP. THEY ALSO HAVE HIGHER HOSTILITY SCORES, PARTICULARLY IN THE SELF-PUNITIVE DIRECTION. THESE SCORES SUPPORT THE IDEA THAT SELF-INJURY IS AN ACT OF AGGRESSION AGAINST THE SELF TO APPEASE GUILT. OTHER FACTORS SUGGEST THAT A CONSCIOUS MANIPULATION OF THE ENVIRONMENT IS ALSO A COMPONENT OF SELF-INJURY. SUPPORTING STATISTICS ARE PRESENTED IN TABULAR AND GRAPHIC FORM. (AUTHOR ABSTRACT MODIFIED).