Accurate estimate of drug exposure plays an important role in studies of the neurobiology of drug dependence. The validity of self-reported drug use by subjects participating in such studies has not been well established. This study examined the relationship between self-reported drug use and biological markers in 18 non-treatment-seeking cocaine-dependent individuals participating in research on the effects of cocaine on the brain. A significant relationship was found between self-reported frequency of cocaine use and hair cocaine concentration. Frequency of alcohol use correlated significantly with plasma carbohydrate-deficient transferrin and aspartate aminotransferase levels. These results suggest that self-reported substance use in non-treatment seeking research subjects is generally valid.