The research is based on an analysis of data collected by Project GANGMILL, in which 2,865 incarcerated offenders in 23 correctional facilities in seven States provided usable responses to a series of 94 questions. One sequences of 7 questions addressed serious misconduct within the prison. A second group of nine questions asked subjects about a variety of unrelated deviant activities in which they may have participated. The research found substantial evidence for a general deviance syndrome; evidence was also found for the effect of the institution on offender behavior. In all, nine predictor variables accounted for 31.8 percent of the variance in prison deviance. A model of deviant behavior was developed as an explanation for the research findings. First, a person's background factors determine the degree of his/her bond to mainstream social values, which in turn influences a general tendency to engage in deviant behavior. It is likely that particular circumstances also contribute to a preference for one or more specific types of deviance, such as a preference for drug abuse (perhaps to self-medicate). These general and specific predispositions to deviate from social norms then interact with a person's situation, which provides both opportunity and instigation to particular deviant acts. The clustering together of inmates from more deviant backgrounds may contribute to an institutional climate that encourages still higher levels of rule-breaking. A "contagion effect" could easily explain the correlation between prison deviance and institutional climate. Although such clustering is often required by internal and perimeter security needs, insofar as is practical, concentrations of offenders with highly deviant backgrounds should be avoided. On the other hand, positive institutional behavior is somewhat predictive of good non-prison behavior and can act as a proxy of in-prison treatment effectiveness until a better outcome measure is available. 3 tables and 25 references