At the inmate level, prison rule infraction data are extensively used for classification purposes and to categorize offenders by degree of security, program assignment, and furlough eligibility. Rule infraction data are also used to predict the likelihood of future prisoner dangerousness and in the granting or withholding of good time and parole. Sources of measurement error in official prison data relate to correctional officer discretion, definitional issues, inmate behavior detection, inmate status characteristics, temporality, and jurisdiction. Given these sources of error and their threat to data validity and reliability, the use of multimethod data collection and comparison strategies is critical. Research should examine prison inmate self-reports, inmate victimization data, and officially gathered infraction measures at each stage of the prison disciplinary process. Research should likewise be conducted on samples of prisoners, correctional officers, and prison administrators in prisons within one correctional system and across correctional systems in multiple jurisdictions. Social scientists should continue to develop theories that explain and predict the behavior of both prison guards and prisoners. 33 references and 5 notes