The diagnosis of depression by clinical interview may be difficult in the patients with communication problems (aphasia, severe cognitive impairment or severe deafness). In these cases, depressive symptoms may be observed by others (nurses and caregivers). The aim of this study was to evaluate the clinical usefulness of an observational scale to identify depression in older patients. Seventy-six institutionalized patients in an intermediate-long-term care center were evaluated. Of them, 39 were excluded because they were unable to perform a clinical interview, needed to diagnose depression. Of the excluded patients, 18 had aphasia, 7 showed severe cognitive impairment: their mini mental state examination (MMSE) score < 14, and 14 collaborated very poorly. Thus 37 patients were analyzed, mean age was 83 +/- 0.86 years (30 women and 7 men). Diagnostic categories were: neurological 16 patients (43.2%), fractures/orthopedics 6 (16.2%), pulmonary/cardiology 5 (13.5%) and others 10 (27.1%). The mean Barthel index was 57.0 +/- 31.6 and mean MMSE score was 21.1 +/- 4.3. The observational scale (OS) designed with six items, was applied to all patients. Each item was scored as never (0 points), sometimes (1 point) and always (2 points). Thus total OS score ranged from 0 to 12. Two observers, who knew the patients (nurses), applied the OS. A trained geriatrician, using the 15-item geriatric depression scale (GDS) performed detection of depressive symptoms. There were 15 patients (40.5 %) with depression on the GDS. OS scale score with a cutoff point of 5 or more was present in 13 patients; nine of them had depression (69.2 %). In the remaining 24 patients with an OS score < 5, depression was present only in 6 cases (25%) (chi2 = 6.844; p < 0.01). The OS>>/= 5, in the present study, obtained a sensitivity of 60 %, a specificity of 81 %, a positive predictive value of 69 %, and a negative predictive value of 75 %.We concluded that (i) the OS has been useful for identifying depressive symptoms with an acceptable sensitivity and specificity, and (ii) the OS may be an alternative to detect depression in patients who are unable to perform a clinical interview.