Deferred prosecution conditionally halts the criminal justice process prior to prosecution, allowing appropriate persons to avoid voluntarily the stigmatizing criminal justice process by choosing to participate in a community-based program targeted at diagnosed client needs. The program is equivalent to pretrial probationary supervision, successful completion of which cancels further criminal justice proceedings. Null hypotheses formulated for the CPA study were (1) CPA referees do not display significantly distinguishing variables from those warranted; (2) the CPA-accepted group does not display significantly distinguishable characteristics from those rejected by CPA; and (3) the CPA-accepted and the warranted shoplifter samples are not significantly different. The study sample was the total population of referrable offender reflected by the January 1, 1973 - January 1, 1974 CPA-warranted and referred file for the specific felonies eligible for referral to CPA. Findings showed a bias to refer the higher education groups, students, women, and whites. Race, however, was the only variable which remained significant when controlled by prior adult record. The shoplifter as a theoretically defined offender type was found in disproportionately large numbers in the accepted sample. A theoretical model is presented as an explanation of and support for the CPA offender premise. Overall, CPA is functioning according to objectives, despite the limitations identified. Recommendations for improvement are suggested, and some legal issues involved in the use of restraints without formal legal proceedings to establish guilt are discussed. Tabular and graphic data and approximately 80 references are provided. Forms used in the CPA program are appended.