BACKGROUND Malnutrition, associated with poor outcome in the elderly, may be exacerbated by weight loss during hospitalization. Accurate estimation of energy requirements is important, particularly if predictions are applied to caloric supply. Because data on energy requirements for the elderly are limited, particularly for African-American patients, predictions are commonly made with equations derived from a younger, caucasian, cohort from 80 years ago. METHODS To compare measured resting metabolic rate (RMR) in a hospitalized elderly African-American cohort in an urban community hospital during 1998 with Harris-Benedict predictions of basal energy expenditure (BEE). Energy expenditure was measured by a strict protocol with a portable metabolic cart, and height and weight were measured standardly and used to calculate BEE and body mass index (BMI). RESULTS In 61 subjects, aged 79.6+/-8.9 years, measured RMR was significantly greater than predicted BEE (p=0.001, t-test). Caloric expenditure averaged 24.7+/-5 kcal/kg/day, but the range was broad (14-39 kcal/kg/day). The BEE prediction was 20.3+/-2.4 kcal/kg/day. CONCLUSIONS In elderly hospitalized African-American patients, the Harris-Benedict equation significantly underestimated energy requirements. Given the link between unintentional weight loss and increased mortality on the one hand and potential clinical complications of overfeeding on the other, measurement of energy expenditure is warranted.