Conditions addressed in the experiment were those assumed to be critical in generating a system of human relationships consistent with either a correctional orientation or the goal of humane containment: respecting the dignity and individuality of inmates, and giving inmates opoportunities to progress towards increasing self-control over their life circumstances in preparation for release. A baseline experimental prison environment emphasizing custody was modeled on medium-high security prisons in Australia. Compared with it were two more liberal environments: and individualized custodial environment, with officers trained to maintain security in a manner that allows inmates to retain their self-respect; and a participatory environment, where the emphasis was on the encouragement of constructive and responsible behavior. Subjects for the 4-day experiment were recruited from newspapers and participated either as prisoners or as officers in surroundings that were meant to simulate real prison conditions. Results affirmed that hostile, affrontive relations in prisons are a function of prison social organization rather than the personal characteristics of prisoners. The standard custodial regime induced ordinary persons with little knowledge and no experience of prisons to behave in much the same way as prisoners and officers in real institutions. Yet changes in the experimental prison regime produced dramatic changes in the relations between officer and prisoner subjects. That is, both officers and prisoners responded to the explicit and implicit definition of the situation provided by their superiors (those with power over them). There seems to be good reason to believe that the behavior of real guards and real prisoners is similarly determined. This investigation remains to be taken further by demonstrating that the same changes produce similar effects in the corresponding real life situation. Comments and comparisons are made regarding the Zimbardo Stanford Prison Study. (Author abstract modified)