The affinity of a monoclonal antibody for its tumor-associated antigen is one of several parameters governing in vivo monoclonal antibody distribution. However, there is a lack of apparent correlation between the affinity of a bivalent monoclonal antibody measured using equilibrium binding experiments and its in vivo delivery. Furthermore, differences in the reported affinity for identical antibody/antigen pairs are quite common in the literature. In this paper, both of these discrepancies are addressed in terms of variation in avidity due to bivalent interaction. The enhancement of avidity afforded by bivalent attachment is addressed theoretically by extending the model of Crothers and Metzger (Immunochemistry, 9: 341-357, 1972). Theoretical assessment of Lineweaver-Burk, Scatchard, Steward-Petty, Langmuir, fluorescence recovery after photobleaching, and Sips models demonstrates quantitatively that the measured affinity using equilibrium binding experiments may vary over four orders of magnitude with similar variation in experimental cellular antigen density. Further, the measured affinity is a function of the experimental protocol. Predictions of avidity enhancement were confirmed experimentally using fluorescence recovery after photobleaching. These experiments measured the equilibrium binding constant and concentration of binding sites for an immunoglobulin G monoclonal antibody and its F(ab) fragment directed against the rabbit VX2 carcinoma cell line. Bivalent binding data agree quantitatively with those predicted by the bivalent model with no adjustable parameters. It is concluded that bivalent equilibrium binding constants are useful only in antibody screening, where experimental conditions are identical for all series. They must be used with caution in predicting in vivo antibody distribution, and it is recommended that the intrinsic, monovalent affinity be measured in tandem with any bivalent antibody study as a standard reference.