Urban, poor, crack cocaine-dependent clients were randomly assigned to outpatient addiction counseling (n=39) or day treatment (n=40). Participants in both conditions received equivalent individual cognitive-behavioral counseling and earned equivalent payment vouchers for providing cocaine-negative urine samples. However, day treatment participants attended significantly more psychoeducational and recreational groups and received two meals per day. Prior to random assignment, more participants expressed a preference for day treatment and participants were more likely to return for an initial appointment following assignment to day treatment. However, no significant between-groups differences in tenure or abstinence were detected during the 3-month course of treatment. These null findings were attributable to an absence of a dose-response effect for the group interventions in the day treatment condition. In addition, there may have been a ceiling effect from the vouchers, which masked the influence of the additional day treatment components. © 2002 Elsevier Science Ltd. All rights reserved.