Heparin-induced thrombocytopenia (HIT) is a potentially serious complication of heparin therapy. Antibodies to platelet factor 4 (PF4)/heparin complexes have been implicated in the pathogenesis of this disorder, but the antigenic epitope(s) on the protein have not been defined. To address this issue, we studied the binding of HIT antibodies to a series of recombinant proteins containing either point mutations in PF4 or chimeras containing various domains of PF4 and the related protein, neutrophil activating peptide-2 (NAP-2). Serum samples from 50 patients with a positive 14C-serotonin release assay (14C-SRA) and a clinical diagnosis of HIT and 20 normal controls were studied. HIT antibodies reacted strongly with wild- type (WT) PF4/heparin complexes, but reacted little, if at all, with NAP- 2/heparin complexes (optical density [OD]405 = 2.5 and 0.2, respectively). Alanine substitutions at three of the four lysine residues implicated in heparin binding, K62, K65, and K66, had little effect on recognition by HIT antibodies (OD405 = 2.2, 2.8, and 2.0, respectively), whereas an alanine substitution at position K61 led to reduced, but still significant binding (OD405 = 1.0). Similar studies involving chimeras between PF4 and NAP-2 localized a major antigenic site to the region between the third and fourth cysteine residues for more than half of the sera tested. This site appears to involve a series of amino acids immediately after the third cysteine residue beginning with P37. Thus our studies suggest that whereas the C-terminal lysine residues of PF4 are important for heparin binding, they do not comprise a critical antigenic site for most HIT antibodies. Rather, we propose that maintaining a region near the third cysteine residue of PF4, distal from the proposed heparin-binding domain, is required to form the epitope recognized by many HIT antibodies.