BACKGROUND & AIMS: Methods for estimation of nutritional expenditures for hospitalized patients may not be sufficiently specific. This study aimed to investigate the accuracy of predictive equations compared to indirect calorimetry (IC) and the effect of certain patient characteristics which might correlate with total daily energy expenditure on a heterogeneous population of hospitalized medical patients. METHODS: A cross sectional study including demographic information, measures of bioelectric impedance analysis (BIA) including height and bodyweight (BW), IC, heart rate and from patient records, information was collected regarding nutritional risk by Nutrition Risk Screening 2002, biomarkers of C-reactive protein (CRP), albumin and leukocytes. The Harris-Benedict (HB), Mifflin St. Jeor (MSJ), and Schofield equations were calculated. Data were analyzed using T-test, linear and logistic regression analysis. RESULTS: Overall, 197 patients, mean age 63.6 ± 16.0 years were measured with IC and had equations performed. BIA was performed in 187 and 46 withdrew, as they were too ill to measure, has oxygen or forgot fasting. All estimation methods underestimate energy expenditures for patients at nutritional risk (p <
0.001), and HB and MSJ underestimate for those with body mass index (BMI) <
18.5 (p = 0.029 and p <
0.001), while for BMI≥30 all overestimate but only HB significantly (p = 0.025). Elevated CRP and leukocytes, lower heart rate, lower and higher BMI, older patients and patients at nutritional risk can affect estimated total daily energy expenditure by equations compared measured by IC (p <
0.05). CONCLUSION: HB, MSJ, and Schofield equations all underestimate energy expenditures with higher variations in patients at nutritional risk. In patients with BMI≥30, energy expenditures are overestimated. Considerations are to measure energy expenditures for patients at nutritional risk with continued weight loss and need for artificial nutrition, and for those with BMI≥30.