THE TRUST BEGAN IN 1972 TO PROVIDE SHELTER AND GROUP COUNSELING TO SMALL GROUPS OF MEN AND WOMEN FROM AGES 17 TO 25. MOST OF THE RESIDENTS WERE AGED 17 TO 19, AND ONLY A FEW WERE OVER AGE 20. A SPECIALLY DEVELOPED GRID AND OTHER TESTS WERE COMPLETED BY 199 RESIDENTS ON ADMISSION AND BY 90 MEN AND 42 WOMEN WHO STAYED AT LEAST 1 MONTH. (OF THESE 30 PERCENT STAYED 2 TO 6 MONTHS AND 30 PERCENT MORE THAN 6 MONTHS.) ATTITUDE CHANGE WAS MEASURED FOR FIVE VARIABLES; POSITIVE OUTCOME WAS SEEN AS CHANGE ON THREE VARIABLES. INITIAL MEASUREMENTS OF AGGRESSION AND DEPENDENCE DID NOT AFFECT POSITIVE OUTCOMES. THE SELF-REFERRED CLIENTS SHOWED NO DIFFERENCE FROM PROBATION CLIENTS IN AGGRESSION AND DEPENDENCY BUT DID SHOW AN UNEXPECTEDLY HIGH POTENTIAL FOR 'DIFFICULT BEHAVIOR.' SCORES WERE COMPARED WITH THOSE OF A MATCHED GROUP OF DETENTION CENTER RESIDENTS. THE TRUST CLIENTS SCORED 30 PERCENT ON THE DIFFICULT BEHAVIOR ITEM, THE DETENTION CENTER RESIDENTS 10 PERCENT. OVER TIME, HOWEVER, A SIGNIFICANTLY LARGER PERCENTAGE OF TRUST RESIDENTS BECAME MORE LAW-ABIDING. THEY ALSO SHOWED GREATER SELF-ESTEEM AND IMPROVED FAMILY RELATIONSHIPS. POSITIVE OUTCOMES WERE SIGNIFICANTLY ASSOCIATED WITH A SENSE OF PURPOSE SHOWED BY THE HOUSE DIRECTOR, THE COHESIVENESS OF THE HOUSE GROUP, AND RELATIONSHIPS FORMED WITHIN THE GROUP. (RELATIONSHIPS WERE ESPECIALLY IMPORTANT FOR FEMALES.) TRANSIENTS DISRUPTED THIS COHESIVENESS. AN ANNEX FOR TRANSIENTS IS RECOMMENDED. DETAILS OF THE STUDY ARE GIVEN. REFERENCES ARE APPENDED.