Sedation may be used in intensive care and emergency medicine to improve the oxygen demand/delivery ratio. The influence of sedation has most frequently been investigated in a dose-related manner. The aim of the present study was to determine the effect-related influence of different sedatives on oxygen uptake (VO2) in relation to defined resting conditions. METHODS. Forty ASA I patients who had to undergo a minor surgical procedure were investigated 1.5 h before surgery at basal energy-expenditure measurement conditions. One of the following substances was given with a preset bolus rate in a double-blind, randomised order until a defined level of sleep or side effects was encountered: propofol (n = 8), midazolam (n = 8), thiopentone (n = 8), sodium chloride (n = 8), and fentanyl (n = 8). The sleep level was defined as sluggish response to a loud voice or tapping on the forearm. The variables VO2, carbon dioxide elimination (VCO2), end tidal CO2 (p(et)CO2), oxygen saturation (SaO2), heart rate, systemic blood pressure, skin temperature, and skin resistance on the sole of the foot were documented on-line on a computer. All variables were compared using differences of averages from 10-min periods before and after sedation during which the VO2 was minimal. RESULTS. The mean VO2 before sedation was 264 +/- 60 ml/min, and the measured energy expenditure did differ by -0.2% (+/- 14%) from mean predicted values using the Harris-Benedict equation. The VO2 was reduced by 15 +/- 2% with propofol, by 12 +/- 8% with midazolam, and by 10 +/- 5% with thiopentone. This was statistically significant compared to placebo treatment, as was the difference between propofol and thiopentone effects. All patients in these groups reached the defined sleep level, which was not achieved by the placebo and fentanyl groups. Placebo treatment changed the VO2 by 0.1% (+/- 2%). Fentanyl increased the VO2 by 5% (+/- 8%), which did not reach significance. In the fentanyl group the bolus application had to be stopped at a p(et)CO2 of 50 mm Hg in all patients. In the propofol, midazolam, and thiopentone groups the phasic changes of skin resistance were reduced to zero and the skin temperature increased from 27 +/- 2 degrees C to 32 +/- 2 degrees C. The fentanyl group showed an increase in changes of skin resistance without changes in temperature. CONCLUSIONS. Sleep induced by propofol, midazolam, or thiopentone to a clinically maximal desirable level in spontaneously breathing patients reduced VO2 by 10% to 15%. This level of sedation did not induce a relevant change in P(et)CO2 or SaO2. The effect of propofol appeared to be the most pronounced and least variable. This may be attributable to a more pronounced reduction in single-organ VO2 or to an undetected difference in level of sedation. Fentanyl did, in contrast to most publications on opioid effects, seem to increase VO2. Underlying mechanisms may be sought in an increased rate-pressure product and sympathetic activity on the basis of hypercapnia and changes in muscle tension.