Little is known about the utility of collateral reports in substantiating self-report for individuals assessed in nonalcoholism treatment contexts. This study examined the concordance of 581 pairs of medical patient and collateral responses to a commonly used alcohol screening instrument, the CAGE Questions, as well as to reports of the patient's drinking consequences and alcohol consumption. Results demonstrated that patient/collateral concordance was marginal, but acceptable, on CAGE cut-off scores and, that similar to reports from alcoholism treatment settings, patients generally reported more drinking consequences than collaterals. Patient and collateral reports of the patient's alcohol consumption did not differ significantly. This pattern of patient and collateral reporting of alcohol consequences and consumption was found for both men and women, as well as for patients with a DSM-III-R diagnosis of alcohol dependence. The findings support the validity of patient self-report on alcoholism screening measures in medical settings. Furthermore, results demonstrated that the addition of collateral reports to information directly obtained from patients only modestly improved the identification of alcohol dependence. The overall findings indicate that alcohol screening can be done effectively and efficiently in medical settings.