EIGHT INMATES IN AN ONTARIO, CANADA, MAXIMUM SECURITY PRISON WERE SELECTED BY LOT FROM 16 INMATE VOLUNTEERS TO SPEND 10 DAYS IN SOLITARY CONFINEMENT. THE REMAINING EIGHT SERVED AS A CONTROL GROUP THAT CONTINUED IN REGULAR CONFINEMENT. ANY DIFFERENCES THAT MIGHT OCCUR IN SELF-IDENTITY PERCEPTIONS BETWEEN MEMBERS OF THE TWO GROUPS WERE MEASURED BY THE REPORTORY GRID TECHNIQUE, WHICH PURPORTS TO MEASURE HOW EVENTS ARE PERCEIVED BY THE SUBJECT AS DEFINED IN TERMS OF A SET OF CORE CONSTRUCTS RELATING TO SELF-IDENTITY; STRESS LEVELS WERE DETERMINED BY MEASURING ADRENOCORTICAL ACTIVITY. THE SUBJECTS WERE ALL IN GOOD HEALTH, NONE WERE DIAGNOSED MENTALLY DEFECTIVE, AND NONE HAD ANY EXTENSIVE PSYCHIATRIC HISTORY. THEY DID NOT DIFFER FROM OTHER INMATES IN THEIR TYPES OF CONVICTIONS OR INSTITUTIONAL HISTORIES. THE EXPERIMENTAL SUBJECTS WERE ALLOWED TO TAKE ONLY CLOTHES AND A BLANKET INTO THEIR DIMLY LIT, 8 FEET BY 4 FEET CELLS. FINDINGS SHOW THAT SELF-IDENTITY RANKINGS INCREASED IN STABILITY FOR CONFINED SUBJECTS IN COMPARISON WITH 'CONTROLS,' AND ADRENOCORTICAL FUNCTION, AS MEASURED BY PLASMA CORTISOL LEVELS, SHOWED SOLITARY CONFINEMENT TO BE NO MORE STRESSFUL THAN NORMAL INSTITITUTIONAL LIFE. THESE RESULTS SUGGEST THAT SOLITARY CONFINEMENT IS LESS 'PUNISHING' THAN THOUGHT BY ITS CRITICS. IT MAY, IN FACT, SERVE TO RELIEVE THE STRESS AND IDENTITY CONFUSIONS OCCASIONED BY THE SOCIAL INTERACTIONS OF REGULAR PRISON LIFE. GRAPHIC DATA AND REFERENCES ARE PROVIDED. (RCB)