Since it has been shown that autoanti-IgE may be mistaken for antiallergen antibodies, thus appearing as pseudo-allergen-specific antibodies, it is crucial to separate true-from pseudo-allergen-specific antibodies and to determine to what extent autoanti-IgE appeared as pseudo-allergen-specific antibodies. For this purpose, human Ig pools were affinity-purified successively on a grass-pollen column and then on an antihuman-IgE column. IgG1-4, IgA, and IgM antibodies that were eluted from the grass-pollen column separated into pseudo- (approximately 30-40%) and true-allergen-specific antibodies that were coretained and not coretained, respectively, with the IgE on the anti-IgE column. Levels of autoanti-IgE were determined in individual plasma samples by surface plasmon resonance and statistically compared to the concentrations of allergen-specific antibodies obtained previously in the same plasma samples. A positive correlation between IgM autoanti-IgE levels and grass-pollen- "specific" IgM concentrations (P < 0.0002), and negative correlations between IgA autoanti-IgE and both IgE anti-grass pollen and IgG2 autoanti-IgE levels (P < 0.03, in both cases) were observed for the first time. This supports the contentions that: (1) autoanti-IgE antibodies appeared as pseudo-grass-pollen-specific antibodies, (2) they hid IgE antibodies when the latter were measured, and (3) they compete with one another in binding IgE. Lastly, a model of large Ig complexes is discussed.