Sevoflurane uptake (Vsevo) can be predicted by the square root of time model or the four-compartment model. However, Vsevo and the effect of cardiac output on anaesthetic uptake have not been quantified clinically. After obtaining IRB approval and informed consent, 34 adult patients received closed-circuit anaesthesia with sevoflurane for 1 h. The end-expired sevoflurane concentration was maintained at 2.6% by infusion of liquid sevoflurane into the breathing system. In a subgroup of 12 patients, cardiac output was measured every 5 min by thermodilution (CO group). The effect of patient characteristics (age, height, weight, body surface area) and cardiac output on Vsevo were determined, and Vsevo was compared with the theoretical models. In the CO group, measured cardiac output was used in the formulae of these models. A two-exponential curve described average Vsevo well: Vsevo (ml liquid) = 0 + 1.62 x (1 - e(-2.3)xt) + 18.1 x (1 - e(-0.0089xt), r2 > 0.999. There was no correlation between Vsevo and patient characteristics, except that Vsevo was greater in patients with a greater cardiac output (r2 = 0.36) and cardiac index (r2 = 0.35). The rate of sevoflurane uptake decreased less than predicted by the square root of time and four-compartment models, even when measured cardiac output was used in the formulae. These findings confirm that the square root of time and four-compartment models do not accurately predict anaesthetic uptake. In addition, uptake of sevoflurane cannot be predicted by patient characteristics but was higher in patients with a higher cardiac output.