In a series of 40 cases of lung cancer, we recorded arterial dilution curves before and after unilateral occlusion of the pulmonary artery (on the pathologic side) following injection of dye into the pulmonary arterial trunk ("arterial curves") and in the capillary wedge position ("capillary curves"). In one case, a large pulmonary collateral circulation (PCC) is obvious, i.e., the dilution curve shows the characteristic shape of an early recirculation before and a normal shape after occlusion. In three cases, the PCC is likely to exist for: 1) decreasing of the capillary curves is abnormally slow during control; 2) transit time of capillary curves is obviously longer during control; 3) the surface under the arterial curve is 9 to 20 % larger during control than during occlusion. This surface undoubtly corresponds to the first dye circulation during pulmonary artery occlusion while, during control, it is increased by the PCC inspite of the exponential shape of the decreasing slope. Therefore the dye dilution method cannot be used to calculate precisely the PCC flow. Practically, in lung cancer measurements of the cardiac flow by dye dilution curves is erroneous about once out of 10 times. This conclusion can be extended to other lung diseases where a PCC may develop.