While the use of criminal histories in criminal justice decisionmaking is a sensitive area, this study assumes that it is legitimate for a prosecutor to possess and act upon a defendant's prior record. A preliminary test showed that college students and prosecutors had no difficulties ranking criminal histories by order of seriousness, indicating that the experiment was feasible. Computer programs were then written to generate simulated criminal histories that tested the following variables for the ability to distinguish the seriousness of one prior record from another: length of record, seriousness of each offense, dispositional information excluding the sentence, and time. In the experiment, 226 attorneys from the Kings County, New York District Attorneys Office rated a set of 30 criminal histories on a seriousness scale of 1 to 5 in spring 1980. The first model presented is concerned only with whether a person was ever arrested or ever convicted of a given crime without the number of times. Arrests for major violent crimes -- homicide, robbery, aggravated assault, and assault but excluding rape -- increased the index while trivial offenses such as possession of marijuana or drunkenness decreased the index. Convictions for homicide, carrying a concealed weapon, robbery, assault, and arson were considered important. While convictions for rape and possession of drugs were important, arrests were not. The second model considers only convictions, but produces the same general order of importance: homicide, rape, robbery, arson, assault, sale of heroin, concealed deadly weapon, and burglary. An operational model is shown, and its results are translated into an index. The paper also discusses shortcomings of the index and areas for future research. A description of the computer program design, tables, and two footnotes are provided.