A lthough intravenous Escherichia coli endotoxin has been used extensively in experimental studies to increase lung endothelial permeability, the effect of E. coli endotoxin on lung epithelial permeability has not been well studied. To examine this issue in sheep, bidirectional movement of protein across the lung epithelial barrier was studied by labeling the vascular space with 131I-albumin and by instilling 3 ml/kg of an isosmolar protein solution with 125I-albumin into the alveoli. E. coli endotoxin was administered according to one of three protocols: intravenous alone (5-500 micrograms/kg), intravenous (5 micrograms/kg) plus low-dose alveolar endotoxin (10 micrograms/kg), and high-dose alveolar endotoxin alone (50-100 micrograms/kg). Alveolar liquid clearance was estimated based on the concentration of the instilled native protein. Sheep were studied for either 4 or 24 h. Although intravenous E. coli endotoxin produced a marked increase in transvascular protein flux and interstitial pulmonary edema, there was no effect on the clearance of either the vascular (131I-albumin) or the alveolar (125I-albumin) protein tracer across the epithelial barrier. High-dose alveolar E. coli endotoxin caused a 10-fold increase in the number of leukocytes, particularly neutrophils, that accumulated in the air spaces. In spite of the marked chemotactic effect of alveolar endotoxin, there was no change in the permeability of the epithelial barrier to the vascular or alveolar protein tracers. Also, alveolar epithelial liquid clearance was normal. Morphologic studies confirmed that the alveolar epithelial barrier was not injured by either intravenous or alveolar E. coli endotoxin. Thus, the alveolar epithelium in sheep is significantly more resistant than the lung endothelium to the injurious effects of E. coli endotoxin.