Results found four measures that were relevant for predicting all three forms of misconduct during both study periods: an inmate's age, prior incarceration, prearrest drug use, and program participation. An inmate's sentence length and work assignment hours were also significant predictors of all three outcomes during one or both study periods, with only one of those effects changing significantly over time (sentence length became a stronger predictor of assaults in 1997). The predictors examined were consistently significant across the three sets of models, although it should be noted that an inmate's education and incarceration for violence were relevant for only one or two of the six models examined. The consistency of several of the inmate-level relationships across different outcomes suggests that some of these factors might be useful to consider when making treatment and placement decisions in confinement facilities for men. For example, the findings for the effects of prearrest drug use and incarceration for violence on both assaults and drug/alcohol offenses during incarceration reinforce the link between violent behavior and substance abuse in the offender population. Such a link is not necessarily refuted by the inverse effect of being incarcerated for drugs on the odds of assaults simply because violent drug offenders are more likely to be incarcerated for the crime of violence, with substance abuse as a possible factor in guiding judges towards sending men to prison, as is the case under some structured sentencing schemes. The link between substance-abuse and violence suggests that treating drug problems may go a long way toward making facilities safer. Data were collected from offenders held in State-operated prisons in correctional facilities; the sample included 9,828 men housed in 204 State facilities for 1991 and 10,022 men in 203 facilities for 1997. Tables, notes, references