Out-of-home treatment for youth with conduct problems is increasing rapidly in this country. Most programs for these youth deliver treatment in a group format and commonly employ some version of a token economy. Despite widespread evidence of effectiveness, a substantial minority of treated youth fail to respond. Participants for this study were 3 youth who were nonresponsive to treatment provided in a family-style residential care program with a comprehensive token economy. Our approach to the "nonresponse" of these youth involved modifications of the frequency and immediacy of their access to the backup rewards earned with tokens. We evaluated the effects of the modifications with a treatment-withdrawal experimental design. Dependent measures included two indices of youth response to treatment: intense behavioral episodes and backup rewards earned. Results showed substantial improvement among these indices during treatment conditions.