First, studies of criminality prediction are examined, and Greenwood's prediction research is considered, as is a companion RAND study by the Chaikens. The Greenwood study is criticized for having a flawed research design, for using possibly unreliable self-report techniques, and for asserting that selective incapacitation strategies are useful in alleviating prison overcrowding. Although the Chaikens' study used the same self-report data, it is praised for being more cautious; the Chaikens identify their research as 'postdiction' research, rather than a true prediction study. Finally, selective incapacitation is concluded to be as useless as other sentencing strategies in reducing crime. The second paper justifies the use of selective incapacitation with two arguments. First a candid, public consideration of offender 'dangerousness' is preferable to the arbitrary and unarticulated assumption upon which current sentencing rests. Second, a more selective determination of who should occupy available prison space can remedy overcrowding. Nonetheless, problems with selective incapacitation, including false positives and racial and class bias, are identified. Finally, a defense of the Greenwood study is presented, and moral and ethical objections to the concept of selective incapacitation are addressed. Included are 193 references.