BACKGROUND: Few data exist on compound A during sevoflurane anesthesia when using closed-circuit conditions and sodalime with modern computer-controlled liquid injection. METHODS: A PhysioFlex apparatus (Dräger, Lübeck, Germany) was connected to an artificial test lung (inflow approximately 160 ml/min carbon dioxide, outflow approximately 200 ml/min, simulating oxygen consumption). Ventilation was set to obtain an end-tidal carbon dioxide partial pressure (Petco2) approximately 40 mmHg. Canister inflow (T degrees in) and outflow (T degrees out) temperatures were measured. Fresh sodalime and charcoal were used. After baseline analysis, sevoflurane concentration was set at 2.1% end-tidal for 120 min. At baseline and at regular intervals thereafter, Petco2, end-tidal sevoflurane, T degrees in, and T degrees out were measured. For inspiratory and expiratory compound A determination, samples of 2-ml gas were taken. These data were compared with those of a classical valve-containing closed-circuit machine. Ten runs were performed in each set-up. RESULTS: Inspired compound A concentrations increased from undetectable to peak at 6.0 (SD 1.3) and 14.3 (SD 2.5) ppm (P < 0.05), and maximal temperature in the upper outflow part of the absorbent canister was 24.3 degrees C (SD 3.6) and 39.8 degrees C (SD 1.2) (P < 0.05) in the PhysioFlex and valve circuit machines, respectively. Differences between the two machines in compound A concentrations and absorbent canister temperature at the inflow and outflow regions were significantly different (P < 0.05) at all times after 5 min. CONCLUSION: Compound A concentrations in the high-flow (70 l/min), closed-circuit PhysioFlex machine were significantly lower than in conventional, valve-based machines during closed-circuit conditions. Lower absorbent temperatures, resulting from the high flow, appear to account for the lower compound A formation.