The study was prompted by the killing of two prison guards in the prison's most secure unit in October 1983. The analysis rests on an examination of documents and studies relating to this case and to prisons in general and on comments from 16 prisoners and the warden. Historical data show that little relationship exists between crime rates and incarceration rates. Instead, the use of imprisonment is a function of social policy and the interests and perceptions of policymakers. Increasingly punitive attitudes and 'scapegoating' of disadvantaged groups in society led to increasingly harsh and repressive criminal justice policies in the middle and late 1970's. In response, reform efforts have compromised on previous positions, and reformers have adopted the widespread view that there is a group of high-risk offenders who have no redeeming features. The prison at Marion opened in 1962 and has become the 'end of the line' for both Federal and State prisoners. The prison's control unit, a type of solitary confinement, is used to isolate and punish recalcitrant and violent inmates. However, the prevalence of physical and verbal assaults by inmates and the deaths of the guards show that increased security and repressive measures do not guarantee safety. The need for an examination of the patterns that have developed at Marion and for the development of alternative models is urgent. In sum, the three lessons Marion learned from the experience are that (1) repression does not work, (2) the concept of maximum security needs reexamination, and (3) prisons cannot be isolated institutions free from outside examination of the conditions of confinement. Illustrations and 14 notes are included.