Patients with acute stroke may become malnourished because of feeding difficulties, so dietary supplementation may improve outcome. We investigated the validity and reliability of an 'eyeball' assessment of nutritional status for use in an international trial of different feeding following acute stroke. One hundred and one inpatients (mean age 75 years), of whom 40 had suffered an acute stroke, were independently observed by two medical students, a doctor and a nurse, and categorised as underweight, normal weight or overweight. Body mass index (BMI) was calculated using height and weight. Sensitivity of the informal assessment for correctly classifying the patient as underweight (BMI < 20 kg/m2), normal (20-30 kg/m2) and overweight (BMI > 30 kg/m2) ranged from 64% to 100% and specificity from 71% to 100%. Interobserver agreement between different pairs of observers was moderate to good. An 'eyeball' assessment of nutritional status is valid and reliable enough for use in a multicentre trial of feeding regimens following acute stroke.