OBJECTIVE: To assess the reliability, validity, and feasibility of a French version of the Health of the Nation Outcome Scales for Elderly People (HoNOS65+). METHOD: Twenty mental health professionals completed the scale for 126 elderly patients. We assessed interrater reliability in inpatient and outpatient settings, together with construct and discriminant validity, as well as concurrent validity compared with several widely used clinical instruments. RESULTS: Excellent agreement among raters was observed in the outpatient subgroup (kappa range 0.77 to 1.00), whereas interrater reliability was modest in the inpatient subgroup (kappa values < 0.55). In cases where the patient-caregiver relationship lasted more than 1 month, the reliability of the scale improved. Construct validity analysis revealed no subscale structure. All items except "self-injury" displayed high concurrent validity (significant Spearman correlations with 8 other questionnaires). In terms of discriminant validity, the "cognitive impairment" (area under the curve [AUC] 0.91) and "depressive symptoms" (AUC 0.79) items showed high sensitivity and specificity values. CONCLUSION: Our results indicate that the French version of HoNOS65+ is a promising clinical assessment tool to evaluate mental disorders in the elderly in outpatient settings. We support its use in routine clinical practice, with the condition that a relatively long and individualized patient-caregiver relationship has been established.