A total of 47 sex offenders incarcerated in the same prison and waiting to participate in a cognitive-behavioral treatment program were randomly assigned to either a positive-oriented track (n=24) or an avoidance-oriented track (n=23). The two groups were matched on offender type, age, risk, and sentence length; and all had the same therapists. The majority of the offenders were serving long sentences for serious sex crimes that involved violence or murder. All participants completed a treatment program that lasted approximately 8 months and included approximately 200 hours of group treatment. The two programs differed only in the features of the relapse prevention (RP) components, which constituted about 25 percent of the total intervention. The positive-oriented RP intervention involved identifying the difference between positive-approach and avoidance goals; artwork and written work to describe each client's view of his "Old Me" (at the time of the offending) and "New Me;" the identification of subgoals to be achieved on the way to the New Me; the identification of factors that might prevent the client from achieving his approach goals; and additional skills development to assist clients in achieving subgoals and primary goals. The avoidance-oriented RP component followed the principles of traditional RP in identifying high-risk thoughts, feelings, situations, and behaviors; listing cues for each risk factor; and generating a plan for managing each risk factor. A relapse-prevention interview measured knowledge of risk factors and the quality of planned strategies for dealing with them. Other elements of the evaluation included completion of risk diaries during the last 2 weeks of treatment, therapist ratings, and a self-esteem questionnaire. Evaluation findings indicate that participants who completed the positive-oriented RP intervention had a greater engagement in treatment as measured by homework compliance and willingness to disclose lapses. Also, therapists rated them to be more genuinely motivated to live life without offending by the end of treatment; however, differences between the two groups in terms of self-esteem change following treatment were not very significant. The findings are discussed in terms of their implications for treatment delivery. 2 tables, 19 references, and an appended example of a "New Me" diary