To study the pathophysiological mechanism of progressive hypoxemia after smoke inhalation injury, alterations in ventilation-perfusion ratio (VA/Q) were studied in an ovine model by using the multiple inert gas elimination technique. Because ethane was detected in expired gas of some sheep, we replaced ethane with krypton, which was a unique application of the multiple inert gas elimination technique when one of the experimental gases is present in the inspirate. Severity-related changes were studied 24 h after injury in control and mild, moderate, and severe inhalation injury groups. Time-related changes were studied in controls and sheep with moderate injury at 6, 12, 24, and 72 h. Arterial PO2 decreased progressively with severity of injury as well as with time. In smoke-exposed animals, blood flow was recruited to low VA/Q compartment (0 < VA/Q < 0.1; 17.6 +/- 10.6% of cardiac output, 24 h, moderate injury) from normal VA/Q compartment (0.1 < VA/Q < 10). However, increases in true shunt (VA/Q = 0; 5.6 +/- 2.5%, 24 h, moderate injury) and dead space were not consistent findings. The VA/Q patterns suggest the primary change in smoke inhalation injury to be a disturbance of ventilation.