Possible causes for the failure of immunoassays to detect anti-acetylcholine receptor activity in serum from confirmed myasthenia gravis (MG) patients were investigated. A more sensitive assay, using Protein A to trap immune complexes (ARIA), was applied to 65 MG sera which were negative in the usual assay and to 42 normal human sera. Normal and negative MG sera had antibody (Ab) activity in the same range (50-70 pM). Titers present in 70% of normal sera appeared to the specific antireceptor antibodies as defined by tests for antigen specificity. Thus, higher sensitivity assays did not improve discrimination of MG from normals. In a second group of 108 MG sera studied, 48 were negative by the usual assay criteria in a rat acetylcholine receptor immunoassay. Further detailed analysis of this negative group showed that 3 48 had IgG3 antibody not detectable in the test, 14 48 had Ab's recognizing human receptor determinants exclusively, 29 48 had toxin blocking Ab's not determined by immunoassays, and 6 48 were negative in all tests. The results indicate that the exclusive occurrence of toxin-blocking antibodies in MG subjects is a major factor contributing to false negatives in the ARIA test. Estimates of the amount of Ab's with this functionality indicated that they are present in very much smaller amounts than other classes of anti-receptor Ab's. Degree of blocking activity in patient serum showed a fair correlation with severity of disease. Thus, blocking antibodies appear capable of causing all degrees of disease severity in the absence of other types of antireceptor Ab's. The development of a sensitive and quantitative in vitro assay for blocking antibodies combined with the usual immunoassay would be a major improvement for a MG diagnostic test, with >94% positivity predicted.