THE INVESTIGATION INVOLVED THE POPULATION OF ALL KNOWN PRISON HOMICIDES IN 1973 WHICH OCCURRED IN ADULT MALE FELON STATE OR FEDERAL INSTITUTIONS HOUSING 200 OR MORE INMATES; THESE HOMICIDES NUMBERED 128. RELEVANT DATA WERE COLLECTED FROM SURVEYS OF LOCAL, STATE, AND FEDERAL AUTHORITIES AS WELL AS FROM EXTENSIVE FIELD WORK BY THE PROJECT STAFF. RESEARCH FINDINGS OF THE 18-MONTH UNDERTAKING SHOWED THAT KILLINGS ARE MOST COMMON AMONG MAXIMUM SECURITY PRISONERS WITH HISTORIES OF VIOLENT OFFENSES AND THAT, CONTRARY TO EXPECTATIONS, GANG CONFLICTS AND RACIAL ANTAGONISMS WERE NOT THE PREDOMINANT FACTORS IN PRISON HOMICIDE. WITH RESPECT TO MORTALITY IN PRISON, IT WAS FOUND THAT THE CRUDE DEATH RATE WAS LOWER THAN THAT OF THE NONINCARCERATED POPULATION, ALTHOUGH BOTH THE SUICIDE AND HOMICIDE RATES WERE HIGH. CONCERNING THE PRISON ENVIRONMENT ITSELF -STAFF, PHYSICAL FACILITIES, AND PENAL PROGRAM -- THERE WAS LITTLE EVIDENCE THAT THE OCCURRENCE OF HOMICIDE IS RELATED TO THE PRESENCE OR ABSENCE OF MODERN REHABILITATIVE INFLUENCES. A CRUCIAL ASPECT OF PRISON HOMICIDE WAS THE DETERMINATION OF THE RELATIONSHIP BETWEEN THE VICTIM AND MURDERER. TWO TYPES OF HOMICIDE WERE IDENTIFIED: THOSE IN WHICH THERE WAS A SINGLE ASSAILANT; AND THOSE IN WHICH THERE WERE TWO OR MORE PRISONERS INVOLVED AS MURDERERS. SIMILARITIES BETWEEN PATTERNS OF HOMICIDE IN PRISON AND IN SOCIETY LIE IN THE CONCENTRATION OF HOMICIDES WITHIN THE VIOLENCE-PRONE SEGMENT OF THE POPULATION (ALL OF THE HOMICIDES REPORTED OCCURRED IN 27 STATES; 23 STATES HAD NO PRISON HOMICIDES). DIFFERENCES LIE IN THE MORE INSTRUMENTAL OR PRECIPITATING FACTORS, SUCH AS LACK OF FIREARMS, UNAVAILABILITY OF ALCOHOL, AND ABSENCE OF FAMILY MEMBERS AS POTENTIAL VICTIMS. TABULAR DATA, A LIST OF RELATED READINGS, AND AN INDEX ARE PROVIDED. (AUTHOR ABSTRACT MODIFIED).