THE EXPERIMENT WAS CONDUCTED WITH 34 SEVERELY AND PROFOUNDLY RETARDED ADULT RESIDENTS OF AN INSTITUTIONAL WARD. THERE HAD BEEN A HIGH INCIDENCE OF STEALING AMONG THE RESIDENTS, PARTICULARLY DURING BETWEEN-MEAL COMMISSARY PERIODS. USUALLY THEFTS WERE BLATANT, WITH ONE RESIDENT GRABBING FOOD FROM ANOTHER RESIDENT. THE OVERCORRECTION PROCEDURE, IN WHICH THE THIEF IS REQUIRED BOTH TO RETURN THE STOLEN ITEM AND TO GIVE THE VICTIM AN ITEM IDENTICAL TO THE STOLEN ITEM, WAS COMPARED WITH THE INSTITUTION'S COMMON PRACTICE OF SIMPLE CORRECTION (RESTITUTION), IN WHICH THE THIEF MUST ONLY RETURN THE STOLEN ITEM. UNDER SIMPLE CORRECTION, AN AVERAGE OF 20 THEFTS PER DAY HAD BEEN OCCURRING. OVERCORRECTION REDUCED THEFTS BY 50 PERCENT ON THE FIRST DAY AND BY 75 PERCENT ON THE SECOND DAY. BY THE FOURTH DAY, THEFTS HAD BEEN ELIMINATED. NO FURTHER STEALING OCCURRED. IT IS POINTED OUT THAT OVERCORRECTION SHOULD BE EFFECTIVE WITH NONRETARDED PERSONS, PARTICULARLY WHEN THE VICTIM IS DISLIKED BY THE THIEF. A GRAPH AND A LIST OF REFERENCES ARE INCLUDED. (LKM)