We have examined a new approach, using fuzzy logic, to the closed-loop feedback control of mechanical ventilation during general anaesthesia. This control system automatically adjusts ventilatory frequency (f) and tidal volume (VT) in order to achieve and maintain the end-tidal carbon dioxide fraction (FE'CO2) at a desired level (set-point). The controller attempts to minimize the deviation of both f and VT per kg body weight from 10 bpm and 10 ml kg-1, respectively, and to maintain the plateau airway pressure within suitable limits. In 30 patients, undergoing various surgical procedures, the fuzzy control mode was compared with human ventilation control. For a set-point of FE'CO2 = 4.5 vol% and during measurement periods of 20 min, accuracy, stability and breathing pattern did not differ significantly between fuzzy logic and manual ventilation control. After step-changes in the set-point of FE'CO2 from 4.5 to 5.5 vol% and vice versa, overshoot and rise time did not differ significantly between the two control modes. We conclude that to achieve and maintain a desired FE'CO2 during routine anaesthesia, fuzzy logic feedback control of mechanical ventilation is a reliable and safe mode of control.