The validity of self-reported crack cocaine use among 131 homeless persons participating in an outpatient substance abuse treatment research demonstration project was assessed by comparing the concordance of self-report and urinalysis results. The subjects were participants in either a Usual Care outpatient program or an Enhanced Care day treatment program that included drug free contingent work therapy and housing. For all subjects across four evaluation points, the false negative classification by self-report (i.e., denied verified use) rate for crack cocaine use was 32.0%. Denied verified use was greater in Usual Care (34.9%) than in Enhanced Care clients (23.7%) and greater at follow-up as compared to treatment entry for all clients. The findings are explained in terms of social desirability and the influence of treatment contingencies and greater accountability specific to the Enhanced Care program. The need for validation of self-reported cocaine use data among homeless persons in settings where contingencies are present and in other drug treatment or research settings is recommended. © 1995.