THE PROJECT SETTING CONSISTED OF TWO OPEN-DORMITORY LIVING UNITS. PROJECT CONDITIONS WERE COMPARED OVER THREE STUDY PERIODS. ONE SET OF COMPARISONS INVOLVED AN 8-MONTH BASELINE PERIOD (MAY TO DECEMBER 1975) AND PHASE I (JANUARY 1976 TO MARCH 1977). A SECOND SET OF COMPARISONS INVOLVED A PHASE I AND PHASE II (JULY 1977 TO SEPTEMBER 1978). DURING PHASE I, ONE LIVING UNIT WAS TO REDUCE ITS AVERAGE WARD POPULATION FROM 50 TO 40 WHILE MAINTAINING A 5-POST COVERAGE; THE OTHER UNIT WAS TO MAINTAIN AN AVERAGE OF 50 WARDS BUT INCREASE STAFFING FROM 5-POST TO 6-POST COVERAGE. DURING PHASE II, THE WARD POPULATION AND STAFFING CONDITIONS WERE TO BE REVERSED BETWEEN THE TWO LIVING UNITS. SEVERAL TYPES OF DATA WERE COLLECTED BY THE PROJECT RESEARCHERS: MONTHLY LIVING UNIT REPORTS, RECORDS OF DISCIPLINARY INCIDENTS, AND QUESTIONNAIRES AND INTERVIEWS RELATING TO THE UNIT'S SOCIAL CLIMATE AND PERCEPTIONS OF PROGRAM IMPACT. RESULTS INDICATE THAT REDUCED LIVING UNIT SIZE, WITHIN THE SPECIFIED RANGE (FROM 47 TO 38 WARDS AS ACTUALLY MAINTAINED IN THE PROJECT) AND FOR THE TYPE OF SETTING CONSIDERED, IS CONDUCIVE TO LESS NEGATIVE AND VIOLENT BEHAVIOR AMONG WARDS, FEWER ESCAPES, FEWER TIME ADDS AND MORE TIME CUTS, AND AN IMPROVEMENT IN WARD-STAFF RELATIONSHIPS. BENEFITS INCLUDE SAVINGS IN BED SPACE AND PROGRAM COSTS. TABLES, FOOTNOTES, AND REFERENCES ARE INCLUDED. TABULAR AND STATISTICAL INFORMATION ARE APPENDED. (WJR)