The isoflurane-saving and CO2-retaining effects of a charcoal filter were compared with a Siemens standard heat and moisture (HME) exchanger and an emptied specimen (dummy). Isoflurane was delivered during the inspiratory phase and consumption investigated at 10, 15 and 25 cycles min−1. The investigation was performed by ventilation with humidified air with a constant end-tidal CO2 and temperature. For a comparison, isoflurane was delivered in a conventional manner via the ventilator. The arrangement with a charcoal filter reduced the isoflurane consumption by a factor of 2.0−2.6, depending on ventilatory rate. Most of the saving was a result of the method of isoflurane delivery (factor 1.4−2.0), while adding the reflector gave a further reduction (factor 1.3−1.5). One circumstance that reduced the net efficiency of the charcoal filter was that it also reflected CO2; consequently, total minute ventilation had to be increased to maintain constant end-tidal CO2.