Specific antibody concentrations are frequently measured in serum (and plasma and intravenous immunoglobulin) samples by enzyme-linked immunosorbent assay (ELISA). The standard negative control involves incubation of buffer alone on antigen-coated wells. The immunoreactivity that develops in antigen-coated wells in which diluted serum has been incubated is assumed to represent specific antibody binding. This approach can result in marked overestimation of specific antibody levels, because serum contains specific polyvalent antibodies which bind, primarily with low affinity, to multiple antigens (including those on ELISA plates) despite the use of blocking agents. Non-denaturing purification of serum IgG, followed by assessment of the antigen binding or antigen-binding affinity of this purified IgG, can reduce but not eliminate the problem of polyvalent antibody binding in indirect ELISAs. Alternatively, polyvalent antibody binding can be estimated by incubating a diluted serum sample on wells coated with an irrelevant protein (such as bovine serum albumin or a scrambled peptide sequence) or buffer alone, then subtracting this reactivity from the sample's binding to wells coated with the antigen of interest. Polyvalent binding of immunoglobulins must be accounted for in order to obtain accurate ELISA measurements of serum, plasma, or intravenous immunoglobulin antibodies.