The various meanings of data are examined, and the importance of data to the scientific attitude is emphasized. Some criminal justice data sources are then analyzed, beginning with the Uniform Crime Reports, data for which are obtained by asking officials, such as police, about crimes known to them (i.e., the number of crimes that have been cleared by arrest, the number of law enforcement officers killed, the value of property stolen and recovered, etc.) Among the limitations of this data source are the number of unreported crimes, changing police practices in crime reporting, inadequacy of clerical staff in reporting, and political pressures to manipulate data should be obtained, if possible, from the FBI which maintains nationwide records based on fingerprints, in preference to State or regional identification centers. Victimization surveys present an incomplete picture of crime because the victims do not always report crimes to the police, perceiving them as often uninterested and unresponsive. Another impediment to the widespread use of victimization survey methods for evaluations in their cost. Other criminal justice data systems include data gethered by the Department of Health and Human Services on causes of death, including homicide and suicide; data systems focusing on persons under correctional supervision; Uniform Parole Reports; the Offender-Based System of Correctional Information and Statistics, and the Offender-Based Tracking System (OBTS). Regarding this last system, some pitfalls are involved in attempting to trace an offender's history through disposition records, often ambiguous in terms of the actual sanction imposed. Overall, the greatest hazard to the criminal justice user is forgetting the operational meanings of the concepts employed. These are indispensable to an evaluator in choosing statistical methods appropriated to the data at hand. Notes and 54 references are appended.