Malnutrition and its impact on clinical outcome may be underestimated in hospitalised elderly as many screening procedures require measurements of weight and height that cannot often be undertaken in sick elderly patients. The ‘Malnutrition Universal Screening Tool’ (‘MUST’)has been developed to screen all adults, even if weight and/or height cannot be measured, enabling more complete information on malnutrition prevalence and its impact on clinical outcome to be obtained. In the present study, 150 consecutively admitted elderly patients (age 85 (sd 5·5) years) were recruited prospectively, screened with ‘MUST’ and clinical outcome recorded. Although only 56% of patients could be weighed, all (n 150) could be screened with ‘MUST’; 58% were at malnutrition risk and these individuals had greater mortality (in-hospital and post-discharge, P<0·01) and longer hospital stays (P=0·02) than those at low risk. Both ‘MUST’ categorisation and component scores (BMI, weight loss, acute disease) were significantly related to mortality (P<0·03). Those patients with no measured or recalled weight (‘MUST’ subjective criteria used) had a greater risk of malnutrition (P<0·002) than those who could be weighed and, within both groups, clinical outcome was worse in those at risk of malnutrition. The present study suggests that ‘MUST’ predicts clinical outcome in hospitalised elderly, in whom malnutrition is common (58%). In those who cannot be weighed, a higher prevalence of malnutrition and associated poorer clinical outcome supports the importance of routine screening with a tool, like ‘MUST’, that can be used to screen all patients.