OBJECTIVE Given the inaccessibility of indirect calorimetry, intensive care units generally use predictive equations or recommendations that are established by international societies to determine energy expenditure. The aim of the present study was to compare the energy expenditure of critically ill patients, as determined using indirect calorimetry, to the values obtained using the Harris-Benedict equation. METHODS A retrospective observational study was conducted at the Intensive Care Unit 1 of the Centro Hospitalar do Porto. The energy requirements of hospitalized critically ill patients as determined using indirect calorimetry were assessed between January 2003 and April 2012. The accuracy (± 10% difference between the measured and estimated values), the mean differences and the limits of agreement were determined for the studied equations. RESULTS Eighty-five patients were assessed using 288 indirect calorimetry measurements. The following energy requirement values were obtained for the different methods: 1,753.98±391.13 kcal/day (24.48 ± 5.95 kcal/kg/day) for indirect calorimetry and 1,504.11 ± 266.99 kcal/day (20.72±2.43 kcal/kg/day) for the Harris-Benedict equation. The equation had a precision of 31.76% with a mean difference of -259.86 kcal/day and limits of agreement between -858.84 and 339.12 kcal/day. Sex (p=0.023), temperature (p=0.009) and body mass index (p<0.001) were found to significantly affect energy expenditure. CONCLUSIONS The Harris-Benedict equation is inaccurate and tends to underestimate energy expenditure. In addition, the Harris-Benedict equation is associated with significant differences between the predicted and true energy expenditure at an individual level.