Strict use of desert theory would mandate that (1) all persons convicted of crimes punishable with imprisonment be imprisoned, (2) persons convicted of serious crimes punishable by imprisonment not be allowed to plea bargain, (3) multiple offenders be given consecutive rather than concurrent sentences, and (4) the importance of harm be substantially reduced in determining sentence. Seemingly, the implementation of desert theory would increase prison populations; ways that it can reduce imprisonment are explored; for example, most offenders in the United States serve more time than do offenders in other countries. Since proportionality is the keystone of desert theory, the severe reduction in actual time served required by a true desert approach may balance the increased time served by the increased number of offenders required to serve time. If overcrowding should result, reduction of prison populations is necessary; this can be achieved by reducing the duration of imprisonment as a matter of percentage of the original sentences. The effect of selective incapacitation on reform of substantive criminal law is considered, and the empirical confusion of selective incapacitation criteria is reported. Use of a desert model is concluded to be preferable to use of selective incapacitation. Finally, in establishing a desert sentencing scheme, it is recommended that guidelines be established to prevent prison overcrowding; further, they should be based upon prediction of likely attachment populations. Eighty-four references are included.