Abstract Hydrocarbon aspiration is a relatively common cause of pulmonary injury in the pediatric age group. We evaluated the pathophysiology of kerosene aspiration in 18 anesthetized dogs during the first hour after injury and then compared the effects of treatments by plasmapheresis (n = 6) and positive end-expiratory pressure (PEEP) of 10 cm H 2O (n = 6) with the results obtained in untreated control dogs (n = 6) over the next four hours. A group of uninjured control dogs (n = 3) confirmed that surgical preparation and saline sham injury did not cause acute lung injury like that due to kerosene. The untreated control dogs with a stable pulmonary capillary wedge pressure (Ppw) of 10 torr received 0.5 mL/kg of intratracheal kerosene. Although Ppw was maintained at 10 torr for five hours post-injury in these dogs, there was a wide range of lung edema (wet lung weight to body weight ratios from 16.4 to 61.1 g/kg; median of 41.4 g/kg). Between one and five hours, extravascular thermal volume (ETV) increased by 19.1 mL/kg, associated with a median increase in venous admixture (Ova/ Qt) of 28.5% and a median reduction in lung compliance of 12.7 mL/cm H 2O. Another group of dogs was similarly injured, but underwent batch plasmapheresis one hour after injury to reduce Ppw to a target value of 5 torr for the remainder of the protocol. Plasmapheresis was associated with a smaller change in ETV (0.8 mL/kg, P < .05 compared with injured animals, Kruskall-Wallis). Gravimetric analysis of edema from lungs excised at five hours confirmed the in vivo findings that reduced Ppw was associated with a lower median wet weight to body weight ratio (20.5 g/kg) than in the untreated group ( P < .06, Kruskall-Wallis). Similarly, progressive increases in Qva/Qt (2.6%, P < .06) and decreases in lung compliance (3.7 mL/cm H 2O, P < .05) were prevented by plasmapheresis. A third group of dogs with transmural Ppw maintained at 10 torr underwent identical injury but were treated from one hour with 10 cm H 2O PEEP. In this group, Qva/Qt was also stabilized (median decrease of 2.5%): but edema formation continued (median ETV increase of 28.5 mL/kg). Cardiac output and oxygen delivery were adequately maintained in both treatment groups so that oxygen consumption was not reduced. We conclude that kerosene aspiration caused pulmonary vascular leaks of varied severity and that plasmapheresis reduced pulmonary edema and its adverse effects on gas exchange and lung mechanics in this model. PEEP also reduced Qva/Qt, but did not reduce edema or lung stiffness and so necessitated the use of higher inflation pressures. Because plasmapheresis did not adversely affect systemic oxygen delivery and consumption, this canine study supports a therapeutic approach to early hydrocarbon pneumonitis of seeking the lowest vascular volume or Ppw associated with an adequate cardiac output and oxygen transport.