Protein-energy undernutrition, or the possibility of its development, has been documented to occur frequently in patients on admission to hospital. Deterioration in nutritional status is known to occur in hospital. In a prospective study of 594 sequential hospital admissions, we aimed to assess the prevalence of undernutrition among patients on admission to two acute teaching hospitals in Dublin, Republic of Ireland using the widely-accepted anthropometric criteria applied in a large study from Dundee, Scotland, UK (McWhirter & Pennington, 1994) and to determine changes in nutritional status in hospital. The mean prevalence of undernutrition (11 %) was considerably lower than was reported from Dundee (40 %). Unintentional weight loss before admission and functional impairment on admission occurred to a similar extent in both centres. Weight loss in hospital occurred in the same proportion of patients, but less frequently among those undernourished on admission to hospital, in Dublin compared with Dundee. The patients found to be undernourished on admission in this study had a mortality rate in hospital (6.5 %) over three times that of the adequately nourished group (2 %). The magnitude of the difference in prevalence of undernutrition between the two centres cannot be explained by ethnicity, case-mix or age distribution. With the secular increase in BMI in the population, the thresholds for classifying patients as undernourished or at risk of nutritional deterioration may need to be reviewed. For clinical use, recent weight loss and functional status may be more appropriate variables to use in the evaluation of nutritional status on admission to hospital.