The multiple inert gas elimination technique (MIGET) is a robust tool to assess both ventilation-perfusion (V'A/Q') distributions and the role of extrapulmonary factors determining arterial oxygenation during spontaneous breathing and in mechanically ventilated patients. Mixed expired gas sampling used in the MIGET is most often obtained from a 10-L mixing box (10L-MB) placed in the expiratory side of the ventilator circuit. Consequently, a considerable increase in the compression volume (Vc) would be expected which, in turn, can give rise to potential errors in the estimation of the effective tidal volume delivered to the patient. The effects of the 10L-MB on the Vc were compared with those produced by a newly designed 1-L, mixing box (IL-MB). At a given peak pressure (Ppeak) within the ventilator circuit, the Vc generated by the 10L-MB was about six-times higher than that produced by the 1L-MB. At a Ppeak =50 cmH2O, the Vc were 377 mL (10L-MB) and 67 mL (1L-MB) (p<0.001). In six patients, the mixed expired partial pressures of the six inert gases simultaneously collected from the two mixing boxes fell on the identity line. V'A/Q' distributions recovered using each of the two mixing boxes were equivalent. With the IL-MB, the effects of different positive end-expiratory pressure levels (0, 6 and 12 cmH2O) on Vc and arterial carbon dioxide tension were negligible. In conclusion, the new 1-L mixing box provides efficient gas mixing and substantially decreases the compression volume. It is, therefore, recommended when studies requiring mixed expired gas are performed in ventilated patients.