To select a risk assessment tool to predict which elderly inpatients will fall, two widely used risk assessment scales (Conley and Stratify) were compared to evaluate their sensibility and specificity. A sample of 2651 patients were included in the survey (1620 evaluated with the Conley; 1181 with the Stratify and 173 with both scales). One hundred and 33 patients fell during the hospital stay (5% of the sample but 1% of all the hospital admissions). Both scales show low values of sensibility (0.69 Conley and 0.20 Stratify) or specificity (0.41 Conley and 0.87 Stratify) with a likely hood ratio of 1.18 and 1.55 respectively. Neither scale is more predictive for stratifying at risk patients. The Conley scale is more sensible in discriminating at risk patients but its use must be accompanied by the clinical assessment of the patients.