This work examined the presence of antibodies reacting with human erythrocytes in horse-derived antivenoms used in the treatment of snakebite envenomations, and assessed the efficacy of various fractionation protocols in the elimination of agglutinating antibodies. A number of antivenoms produced by various fractionation protocols were tested for direct agglutination of human erythrocytes. Reactions were observed visually and microscopically, and an indirect anti-equine globulin test was also used. In addition, rabbits and mice were injected intravenously with antivenoms to observe possible intravascular hemolysis and erythrocyte sequestration. All tested antivenoms agglutinated human erythrocytes, albeit to different extent, and also gave a positive anti-globulin test. Agglutination was due to IgG(T) subclass of antibodies. Pepsin digestion of horse IgG, to obtain F(ab')(2) fragments, reduced the direct agglutination, but not the indirect anti-globulin test. Ion-exchange chromatography of IgG in a strongly basic quaternary ammonium cellulose membrane abrogated direct agglutination and reduced the indirect anti-globulin test. Binding of antivenom antibodies to erythrocytes in vivo was demonstrated in rabbits, although there was no evidence of intravascular hemolysis or erythrocyte sequestration in rabbits and mice. It is concluded that anti-human erythrocyte antibodies are present in horse-derived antivenoms, and that fractionation of horse plasma by pepsin digestion, and especially by anion-exchange chromatography, reduces the titer of these antibodies. Our in vivo experimental results do not support a role for these antibodies in early adverse reactions occurring after antivenom administration.