This study validated a French language version of an inventory designed to detect symptoms of depression and anxiety [Goldberg et al, 1987] in a sample of elderly French-speaking inpatients at risk for one of these disorders. Latent trait analysis was used to replicate the structure of the symptoms in the inventory, and receiver operating characteristic analysis was used to assess the performance of the inventory as a screening measure for Major Depressive Episode and Generalised Anxiety Disorder according to DSM-IV criteria. Reflecting the ascertainment of individuals in the sample as being at risk for a disorder, prevalence of individual symptoms was high although the general structure of the inventory was found to be comparable to that found in samples of both community elderly and younger medical patients. ROC analyses showed that the subscales of inventory performed satisfactorily as screening measures for anxiety or depression but lacked specificity for each disorder. In addition to providing further evidence for the utility of this inventory to detect general psychiatric distress in elderly persons, this study provides a valid means of detecting symptoms of depression and anxiety in French speaking groups.