Colloid osmotic pressure (COP) has been used as a predictor of fluid egress from the vascular space and edema development in the lungs and elsewhere. We investigated the relative safety, as predicted by the COP, of 5 percent albumin plus saline and 5 percent albumin alone as replacement fluids during plasma exchange. Fifty-three one and one-half plasma volume exchanges were performed in 15 patients with a variety of diagnoses using intermittent flow cell separators. On specimens obtained, before, during, and after each plasma exchange, the COP was measured directly with a membrane oncometer , and total protein, albumin, and protein electrophoresis were determined using standard biochemical techniques. COP dropped significantly with one to one 5 percent albumin replacement but even more with one-half 5 percent albumin and one-half saline replacement during the exchange. COP did not fall below 12.5 torr at the end of the plasma exchange, even with one-half saline replacement, compared to the 10 to 12 torr level at which pulmonary edema might be expected. The fall in COP during exchange when saline was given first did approach this range but rapidly reversed itself with albumin administration. Clinically, no evidence of tissue or pulmonary edema was observed. Recovery in total protein and COP between plasma exchanges was significant.