Three-strikes laws and other innovations in sentencing policy have revealed a renewed interest in the idea of selective incapacitation. The Rand report's release stimulated much discussion due to the claims that selective incapacitation could reduce crime rates and prison populations simultaneously. Ethical problems inherent in such proposals as well as methodological inconsistencies in the original research led to this re-examination of the proposal and of the empirical basis for the report's conclusions. The current research focused specifically on the methodological issues concerning the construction of the predictive scale. Results revealed that the selective incapacitation scheme advocated by Greenwood and Abrahamse performs extremely poorly in terms of both reliability and validity, thus precluding the implementation of such schemes. However, researchers generally agree that the Greenwood and Abrahamse prediction instrument does very well in accurately identifying low-rate offenders. Therefore, it has been suggested that the best use of a predictive instrument such as this is in making release decisions, as when courts mandate emergency release programs to reduce prison overcrowding. In addition, the use of identification instruments as a tool for evaluating the impacts of correctional policies does not involve predictions of future behavior and thus is less susceptible to ethical challenges. Finally, this study's findings underscored the limitations of incarceration-based crime control strategies. Tables, footnotes, and 70 references (Author abstract modified)