Starling's hypothesis of forces governing fluid movement across capillary membranes suggests that any unopposed decrease in intracapillary colloid oncotic pressure (COP) or increase in capillary permeability should result in increased interstitial fluid. Iso-oncotic increase in pulmonary artery wedge (PAW) causes pulmonary dysfunction. Isobaric reduction of COP with normal capillary permeability does not result in pulmonary interstitial edema. Because sepsis is a frequent antecedent of clinical pulmonary dysfunction, the question was asked: does reduction in the COP-PAW gradient in the presence of sepsis result in increased pulmonary dysfunction? Twenty baboons were studied: group 1--control, group 2--4-h constant infusion of E. coli, group 3--plasmapheresis alone, group 4--plasmapheresis followed by sepsis. Ringer's lactate was infused to keep PAW constant. Arterial and mixed venous blood gases were drawn and the intrapulmonary shunt (QS/QT) was calculated. The data were compared using Tukey's HSD test and one way analysis of variance. Plasmapheresis alone resulted in a 68% reduction in COP (15 +/- 2.9 (SD) torr to 4.6 +/- 0.6 in group 3 and 16.5 +/- 4.3 to 5.7 +/- 0.9 in group 4, p less than 0.05). Sepsis resulted in an increase in QS/QT in all septic animals. There was no increased QS/QT in those animals that had sepsis added to plasmapheresis, group 4 (p less than 0.05). These data indicate that sepsis leads to pulmonary dysfunction but that this dysfunction appears to be independent of colloid oncotic forces.