Abstract Although Lp(a) is an independent risk factor for cardiovascular diseases in humans, the precise pathogenetic mechanisms are still unknown. We have shown that Lp(a) accumulates in human atherosclerotic lesions, and some particles undergo oxidation. Since, following agarose electrophoresis of both plaque extracts and plasma, a region close to the origin immunostained intensely for apo(a) but was lipid-free, we sought to identify whether such samples contained lipid-free apo(a), as previously reported to occur in plaque extracts. Immunochemically identifiable apo(a) was found following density-gradient ultracentrifugation both in the 1.05 < d < 1.09 and the d > 1.21 density fraction from both plasma and plaque extracts. However, because in a competitive binding RIA, displacement curves of apo(a) in plasma and the d > 1.21 were not parallel, it is premature to ascribe a relative amount of total apo(a) to this fraction. Whereas apo(a) immunoblots of SDS-PAGE under reducing conditions of the d > 1.21 fraction of a plaque extract with high apo(a) content showed high molecular weight bands consistent with apo(a) isoforms, the corresponding d > 1.21 fraction showed multiple low molecular weight bands characteristic of fragmentation. Since the d > 1.21 of arterial extracts contained all the material immunostaining for apo(a) migrating towards the cathode, characteristic of immunoglobulins (IgG), we asked whether fragments of apo(a) might have associated with human IgG both in plasma and tissue extracts, or whether our anti-apo(a) reacted with epitopes on human IgG. Immunoblotting with our antiapo(a) of samples of plasma and plaque extracts run on agarose electrophoresis or SDS-PAGE further demonstrated intense staining of multiple bands in the molecular weight range of human IgG. Furthermore, a fraction of plasma and tissue extracts that bound to a protein G affinity column demonstrated immunostaining for apo(a) and was in the size range of IgG. Although one polyclonal anti-apo(a) provided by another laboratory showed the same findings as our antibody, two other polyclonal anti-apo(a) failed to demonstrate immunostaining of human IgG, either on agarose electrophoresis or SDS-PAGE. We speculate that the Lp(a) immunogen used to prepare our anti-apo(a) may have undergone modest oxidation, thus exposing epitopes not normally expressed on apo(a) in native Lp(a). Either antibodies to these epitopes could be recognizing apo(a) fragments, possibly released during oxidation, which are then covalently bound to IgG, or oxidation of apo(a) creates epitopes on apo(a) that are homologous with IgG, thereby leading to cross-reactivity with IgG. Such a scenario could be consistent with recently published studies showing recognition of oxidized LDL by the Fc receptor on macrophages.