Intermittent injection of liquid anaesthetic into a closed breathing system is particularly suitable in countries with limited resources. A method of calculating appropriate times and magnitudes of the injected doses was described by Lowe but the method has never been assessed rigorously. Such an assessment was the purpose of this study. The technique was used in a double-blind, randomized comparison of halothane, enflurane and isoflurane in oxygen-air, with 20 ASA I or II patients in each group, undergoing superficial or abdominal surgery. The prescribed times of injection were adhered to, but the doses, after the first two, were adjusted to maintain systolic arterial pressure within 20% of the reference preoperative value. Partial pressures of the anaesthetics were monitored but concealed from the investigator-anaesthetist. The mean doses found necessary for each anaesthetic were within 33% of those calculated to produce 1.3 MAC. However, end-tidal partial pressure (just before each dose) stabilized at a steady level of only 0.97, 0.42 and 0.77 MAC for halothane, enflurane and isoflurane, respectively. Recovery from enflurane was much more rapid than that from the other agents but no patient admitted to any dreams. We conclude that the rate of uptake of anaesthetic declines more slowly than predicted and that the patients receiving enflurane were less deeply anaesthetized because the greater hypotensive effect of enflurane led to the use of smaller doses.