During the past several years clinical protein A perfusion has attracted much attention because it allows to selectively remove IgG subclasses 1, 2, 4 and probably IgG-containing immune complexes, and has a tumoricidal effect in experimental animals and in some cancer patients. Due to several drawbacks, this therapy is not yet generally accepted. Our first experience with laboratory and clinical protein A perfusions confirms several limitations of this new apheresis therapy. Plasma IgG extraction in the ex vivo system under investigation and during clinical application of protein A perfusion reached a 10:1 ratio of grams IgG removed per gram solid-phase protein A only in 2 of 5 runs. Nevertheless, the absolute amount of IgG removed was very low in all runs due to the restricted protein A load (maximum 200 mg) per column. Removal capacity can be increased by a two-column switchover system with subsequent perfusion and elution. Furthermore, the side effects observed in both in vivo treatments exceeded by far those of other extracorporeal therapies and had not been observed in more than 1,200 unselective plasma exchanges or in 50 cascade filtrations in our center. C3a generation in protein A perfusion is, however, comparable to cascade filtration, but exceeds that of unselective plasma exchange and is lower than in hemodialysis. Consequently, side effects in protein A perfusion cannot be correlated with the total amount of anaphylatoxin generated but may be due to a leakage of protein A or contaminants. Clinical application of protein A perfusion needs a more detailed elaboration in respect to biocompatibility, removal capacity, and the significance of the induced biological effects.