We compared the effects of humidity delivered by the circle system at low fresh gas flows (FGF) with a conventional two-limb and coaxial circuit on the structure and function of the tracheobronchial epithelium in dogs. Animals were anaesthetized and mechanically ventilated using an anaesthesia ventilator to maintain normocarbia. Group I (control) animals received a FGF equal to the required minute ventilation mimicking an open circuit technique. Group II and III animals had FGF set at 20% of the required minute ventilation. Group II used a two-limb circuit and Group III used a coaxial circuit. Relative humidity and temperature of inspired gases were measured at baseline and hourly afterwards. In the first experiment, biopsies of the tracheobronchial tree were obtained bronchoscopically at baseline and then hourly for six hours. Microscopic examination of these samples allowed calculation of mean ciliary length. In the second experiment, tracheal mucus flow velocity (TMFV) was measured at baseline and hourly afterward, using a cinebroncho-fibrescopic method. Delivered absolute humidity was greatest with low FGF and the coaxial circuit, followed by low FGF and a conventional circuit, and high FGF (15 +/- 1.4 vs 9 +/- 0.8 vs 5 +/- 0.4 mg H2O, P < 0.01) after two hours. Mean cilia length (micron) and TMFV (mm/min) fell during the first hour in all three groups. At hour two TMFV returned to baseline in Group III and was significantly greater than Groups I and II (0.8 +/- 0.4 vs 8.6 +/- 1.1 vs 15.4 +/- 2.1, P < 0.001). Mean ciliary length demonstrated a similar pattern with reductions from baseline in all three groups for the first two hours. Groups II and III had an increase in cilia length beginning at hour three and were both significantly greater than Group I at hours 3 through 6 (1.3 +/- 0.5 vs 3.2 +/- 1.1 vs 4.2 +/- 0.8, P < 0.001). Alterations in tracheobronchial structure and function result from exposure to dry gases and are amplified by the duration of exposure. Our findings suggest a minimum of 12 to 15 mg H2O/l is necessary to prevent these alterations. In this study, the combination of low FGF and a coaxial anaesthesia circuit reached this minimum threshold more quickly than a conventional two-limb circuit.