SENTENCED INMATES INCARCERATED IN THE THREE INSTITUTIONS FOR SPECIFIED PERIODS BETWEEN JANUARY 1970 AND JULY 1973 SERVED AS SUBJECTS FOR ANALYSIS. POLICY CHANGES AT ALL INSTITUTIONS IN THE WAY INMATES WERE DISCIPLINED FOR OFFENSES AGAINST PRISON REGULATIONS WERE OCCASIONED BY THE APPOINTMENT OF DIFFERENT SUPERINTENDENTS DURING THE EXPERIMENTAL PERIOD. OFFENSES AGAINST PRISON REGULATIONS WERE WRITTEN IN THE RECORDS OF INMATES AT THE TIME OF THE OFFENSE. DEPENDENT VARIABLES EXTRACTED FROM THE RECORDS WERE THE NUMBER OF OFFENSES OCCURRING IN EACH OF FIVE CATEGORIES AND PENALTIES IMPOSED FOR THEM: FIGHTING, DAMAGING PROPERTY, ABUSIVE CONDUCT, DISOBEYING AN ORDER, AND MISCELLANEOUS. IN ORDER TO QUANTIFY AND COMPARE COMBINATIONS OF SANCTIONS GIVEN OUT BY SUPERINTENDENTS, AVERAGE PENALTIES FOR A PARTICULAR OFFENSE WERE CALCULATED BY DIVIDING TOTAL PENALTIES BY TOTAL OFFENSES IN EACH CATEGORY. ONE OF THE MOST OBVIOUS CHANGES INSTITUTED BY SUPERINTENDENTS WAS IN THE AREA OF SENTENCING INMATES FOR INTERNAL OFFENSES. IN VIEW OF THE DEGREE OF CHANGE IN BOTH SENTENCING POLICY AND FREQUENCY OF OCCURRENCE OF OFFENSES. SENTENCING METHOD IS SEEN AS AN IMPORTANT BEHAVIOR CHANGE AGENT. THE USE OF A PARTICULAR SANCTION IN DIFFERENT INSTITUTIONS HAD VARYING EFFECTS ON BEHAVIOR. ALTERNATIVE EXPLANATIONS FOR THE RESULTS ARE PRESENTED, AND IT IS SUGGESTED THAT CONTINUOUS MONITORING OF DECISIONS AND THEIR EFFECTS BY SUPERINTENDENTS MAY PROVIDE AN EMPIRICAL BASIS FOR THE EFFECTIVE USE OF CONTROL PROCEDURES. SUPPORTING DATA AND REFERENCES ARE INCLUDED. (DP)