The osmotic reflection coefficient (sigma) can be estimated from the increases in hematocrit and plasma protein concentration that result from fluid filtration occurring in an isolated perfused organ. We determined what effect perfusion pump-induced hemolysis has on the value of sigma determined by this technique in both the isolated canine left lower lung lobe (LLL) and forelimb by comparing estimates of sigma obtained before and after correction for hemolysis. Hemolysis was corrected by using the slopes of the relationships between hematocrit and plasma hemoglobin concentration and between the plasma protein and hemoglobin concentrations to correct hematocrit and protein concentration to a state of zero hemolysis. Uncorrected estimates of sigma in the LLL were 1.19 +/- 0.14 (SE) at a venous pressure (Pv) of 12 Torr (n = 7) and 0.90 +/- 0.02 at a Pv of 19 Torr (n = 6). Both sets of LLL's yielded sigma values of 0.77 +/- 0.03 after hemolysis correction. In the forelimb (n = 5), uncorrected and corrected estimates of sigma of 0.99 +/- 0.03 and 0.85 +/- 0.01, respectively, were obtained. The latter values were similar to sigma's (0.88 +/- 0.01) determined by lymph analysis in five additional forelimbs. We conclude that hemolysis results in overestimates of sigma. After hemolysis correction, this technique yields similar results to those obtained from lymph analysis for the forelimb and from published values for the LLL.