Impaired nutritional status has been frequently reported in surveys estimating its prevalence amongst patients in hospital. While there is no doubt that protein–energy undernutrition has serious implications for health, recovery from illness or surgery and hospital costs, lack of nationally or internationally accepted cut-off points and guidelines for most nutrition-related variables make nutritional assessment difficult and proper comparisons between studies impossible. In reviewing published work in which the prevalence of undernutrition has been assessed, it can be seen that each study defined undernutrition, or nutritional risk, using different methodology. This present review aims to highlight the problems which arise when deciphering these studies, and the resulting difficulty in determining the true prevalence of undernutrition and nutritional risk, amongst both general and specific groups of hospital in-patients. It is widely agreed that routine hospital practices can further adversely affect the nutritional status of sick patients in hospital. How this occurs, and the potential effects of impaired nutritional status on clinical outcome are examined. The methods currently available to assess nutritional status are evaluated in the knowledge that such assessments are difficult in clinical practice. The review concludes by proposing that if we want the medical and nursing professions to consider the nutritional status of hospital patients seriously, definitions of undernutrition and nutritional risk, and cut-off values for the nutritional variables measured must be agreed to allow evidence-based practice. Outcome measures which allow clear comparisons between groups and treatments must be used in studies assessing the effects of nutritional interventions.