OBJECTIVE To evaluate the usefulness of the modified Harris-Benedict formula predicting Energy Expenditure (EE) in pressure support ventilated, critically ill patient. METHODS The intensive care unit (ICU) of a teaching hospital. METHODS Fiftyfive patients admitted to ICU for acute respiratory failure, requiring mechanical ventilation. RESULTS After 12 hours at rest, EE was measured using indirect calorimetry (Datex-Deltatrac, with method exclusions - ICEE), and calculated using modified Harris-Benedict formula (MHBEE) (with correction for "hospital activity" and "stress factor") to calculate the bias between calculated and measured EE. Patients were divided into three groups on the basis of nutritional stress: A) non surgical/non septic patients (n = 10), B) complicated surgical patients (n = 21), C) severe infectious/multiple trauma patients (n = 24). In each group, a good correlation between calculated and measured EE was found [A) r = 0.809, p = 0.0046; B) r = 0.753 p = 0.0001; C) r = 0.711, p = 0.0001]. The bias (+/- SEM) was: A 175.1 (+/- 82) kcal/day, B 324.5 (+/- 64.5) kcal/day, C 366.7 (+/- 62.9) kcal/day. The mean difference value seems to be increased in the more stressed patients but these differences did not reach statistical significance (p = 0.23). A single correction factor for the original Harris-Benedict formula (OHBEE) was also calculated (ICEE/OHBEE) on each studied group: A) 1.20 (+/- 0.04), B) 1.28 (+/- 0.03), C) 1.50 (+/- 0.04) (p = 0.0001). CONCLUSIONS The use of both "stress" and "activity" correction factors seems to be excessive in pressure support ventilated ICU patients. A single correction factor, proportional to the intensity of the illness, should be used in mechanically ventilated patients. Compared to the original Harris-Benedict formula, we found an EE increment of about 20%, 30%, and 50% respectively in non-septic/non-complicated, surgical complicated, and multiple trauma/septic patients.