Cocaine-abusing patients randomly assigned to day-hospital or inpatient rehabilitation were compared with patients who self-selected these treatment settings to examine differences in substance use and psychosocial outcomes under experimental and nonexperimental designs. There was little evidence of setting or assignment effects or Setting x Assignment interactions over the 12-month follow-up period. However, Assignment x Time interactions were obtained with 2 measures of cocaine use and measures of family-social and psychiatric problem severity. These interactions indicated greater problem severity at intake among the randomized patients coupled with greater improvements by the 3-month follow-up relative to the nonrandomized patients. Findings suggest that randomized studies of treatment for cocaine abuse may produce somewhat larger estimates of improvement than what is observed in more typical treatment situations.