Little quantitative research has been conducted in this area with the exception of the attention given to prison crowding. Many criminologists have theorized about the effect of prison conditions on inmate behavior within prison and after release. The primary question of the study is whether prison management, through the means at its disposal, creates an environment within a prison that exerts an influence on inmate misconduct that is independent of the individual or collective characteristics of the inmates and staff that live and work at the prison. The data collected covered 121,051 inmates that had been sentenced to Federal prisons and that were incarcerated during June 2001 in the Bureau of Prisons (BOP). BOP electronic records provided information on inmates in such areas as educational and psychological programming, medical care, disciplinary infractions, and movement. The current BOP technique for evaluating institutions’ performance with respect to inmate misconduct was compared with a method that adjusts for the types of individuals housed in the facilities and aggregate characteristics of the prisons. A multilevel approach was invoked that accounts for the nesting of inmates within different prisons. The results show that prisons do differ in their effect on individual inmates’ propensity to become involved in misconduct. Prison management and operations do make a difference in a smaller context than generating riots or reforms. This finding means that there is a capability to use operational data to identify problematic prisons before more catastrophic outcomes become manifest. The study demonstrates that model specification makes a difference in the understanding of which variables are related to misconduct, that the type of misconduct is important for understanding the effects of covariates of misconduct, and that the results of multilevel models can easily be used to compare the performance of prisons. 4 figures, 4 tables, 13 footnotes, 41 references, appendix