The subjects in Rand's second inmate survey were selected to represent a typical cohort of incoming inmates in California, Michigan, and Texas. They completed detailed self-report questionnaires about the variety and intensity of their criminal activity. The reanalysis focused on robbery and burglary only The reanalysis showed the most important contribution of the Rand survey to be its underscoring of the extreme skewness of the distribution of individual offending patterns. Half the offenders commit five or fewer crimes a year, while a small but important group may commit several hundred crimes a year. Estimates of the distribution are sensitive to the treatment of missing data and other factors. Offenders' truthfulness may also affect the observed distribution. Greenwood's prediction scale developed from the data has several problems. Its usefulness varies from State to State, it does not appear to work better than other scales, and it more effectively identifies low-rate than high-rate offenders. Jurisdictions should develop and validate their own scales. Further research is also needed on the operational effectiveness of prediction instruments that might form the basis of efforts to reduce crime through selective incapacitation. Footnotes, data tables, figure, appendixes, and 42 references.