Background Elderly people are at great risk of poor nutritional status. Identifying those at nutritional risk is complex owing to its multifactorial nature and highlights the need for an easily administered nutritional screening tool.Aim To test the robustness of a tool used to identify those patients who require further dietetic assessment.Method The tool was tested on an ‘acute elderly’ ward to assess its statistical robustness, i.e. test–retest reliability, inter-rater reliability, internal consistency and factor structure, concurrent and construct validity. The research was carried out by a dietitian and two nurses on an overall sample of 185 consecutive admissions.Results Test–retest results reflect that the tool is capable of yielding consistent assessments over time. Inter-rater reliability results revealed that when comparisons of agreement between nurses and dietitians were examined there was consistent disagreement. Internal consistency was found to be low but the rotated factor analysis suggested a relationship between the items in the tool. Concurrent validity was exhibited with body mass index, albumin and American Nutrition Risk Index only. Construct validity was not sufficiently exhibited with any of the measures used.Conclusion The tool has not demonstrated reliability as determined by the statistical measures and hence validity at this stage. Issues which need to be addressed to allow this tool to be used appropriately are discussed.