The present study examined the agreement between two measures of prevalence of drug use in the community: self-report and specific pharmacological analyses of urine samples. The data were collected in the context of a random community survey of health practices and attitudes. A random 20% of the households participating in the health study were targeted for biochemical assay. Compliance with urine delivery was relatively high at 79%. Urine samples were screened qualitatively for cannabinoids and benzodiazepines using the enzyme multiplied immunoassay technique (EMIT) (Syva Diagnostics, Palo Alto, CA). Screening for pharmaceuticals used a standard thin-layer chromatography (TLC) technique. Agreement between the self-report and biochemical assay estimates of prevalence was statistically significant (p < .05). While self-report of substance use is not a perfect measure of consumption, it remains a relatively economical and reasonably accurate method of obtaining estimates of substance use in community samples.