If a small number of high-rate offenders commit a disproportionately large amount of crime, targeting limited prison resources on these offenders should achieve increased crime control without increasing prison populations unreasonably. Such a policy depends on identifying high-rate offenders prospectively and early in their careers. Recent efforts to use predictions of individual crime rates as a basis for selective incapacitation are plagued by ethical and empirical problems. An alternative approach relies on observed differences in the criminal career patterns of offenders charged with different offenses to identify offenses committed by offenders who on the average commit crimes at higher rates and have longer careers. This approach may achieve differential incapacitation effects while avoiding many of the ethical and empirical problems posed by sentencing policies that use individual-level predictions to distinguish among offenders. Footnotes, figures, tables, prediction equations, and about 70 references are supplied. (Author abstract modified)