The efficient use of limited prison space could be enhanced by restricting incarceration to those offenders most likely to commit a large number of serious crimes in the future. Selective incapacitation is likely to withstand philosophical and legal challenges only if an effective instrument could distinguish between short-term and long-term offenders and only if prison terms assigned projected long-term offenders do not exceed the reasonable length appropriate for the crime of conviction. Greenwood and Chaikens claim to have developed instruments that identify projected crime rates among offenders. Their analyses are based on identical survey data. Although four of the variables, juvenile and adult convictions and incarcerations, can be identified in official records, errors in such data prevent their fair and effective use. Until the quality of such records is uniformly improved, legal challenges to predictive scores are likely to be successful on the ground of invalidity. The timeliness of such information is also crucial, since decisions bearing upon selective incapacitation must be available at various decisionmaking stages of case processing. 5 footnotes.