Myasthenia gravis (MG) is usually caused by autoantibodies against muscle nicotinic acetylcholine receptor (AChR), which is composed of five subunits (α2βγδ or α2βɛδ). Current treatments, including plasmapheresis, are nonspecific, causing several side effects. We aim to develop an antigen-specific alternative to plasmapheresis, since the latter removes indispensable plasma components in addition to anti-AChR antibodies. We are developing a method for the selective depletion of the anti-AChR autoantibodies from patients' plasma through the construction of “immunoadsorbent” columns carrying AChR domains. We have expressed the extracellular domains (ECDs, amino acids ∼1–210/220) of all human muscle AChR subunits in Pichia pastoris and, in preliminary experiments, in E. coli. The ECDs were immobilized (individually or mixed) on Sepharose beads, producing Sepharose-ECD columns, which were tested for their immunoadsorbing capacity on MG sera and shown to specifically eliminate major autoantibody fractions from several MG sera. The immobilized ECDs remained stable and did not dissociate from their matrix after incubation with serum, whereas the procedure was neither toxic nor immunogenic in two experimental rabbits. Testing the intact or antibody-depleted MG sera and the affinity purified autoantibodies showed that both the intact sera and the purified autoantibodies, but not the antibody-depleted sera, could induce AChR loss in cell cultures and experimental MG in rats. This preliminary study suggests that the myasthenic potency of MG sera is entirely due to their anti-AChR antibodies and therefore their depletion should be of therapeutic value. We conclude that ECD-mediated immunoadsorption can be used as an efficient, antigen-specific therapy for MG.