Monoclonal antibodies (MAbs) often distribute nonuniformly in tumors. In part, that observation reflects intrinsic heterogeneity within the tumor; in part, it reflects poor penetration through tumor substance. Several years ago, we proposed the "binding site barrier" hypothesis (J.N. Weinstein, R.R. Eger, D.G. Covell, C.D.V. Black, J. Mulshine, J.A. Carrasquillo, S.M. Larson, and A.M. Keenan, Ann. NY Acad. Sci., 507: 199-210, 1987; K. Fujimori, D.C. Covell, J.E. Fletcher, and J.N. Weinstein, Cancer Res., 49: 5656-5663, 1989), the idea that antibodies (and other ligands) could be prevented from penetrating tumors by the very fact of their successful binding to target antigen. Calculations suggested that this might be a significant factor in the therapy of even microscopic nodules. The higher the affinity and the higher the antigen density, the greater the barrier. Here, we provide direct experimental evidence of such a barrier to the percolation of D3 MAb through intradermally implanted line 10 carcinoma of guinea pigs. After affinity purification using glutaraldehyde-fixed line 10 cells, the D3 had an average immunoreactivity of 88%, a binding constant of 1.6 +/- 0.3 (SEM) x 10(10) M-1, and saturation binding of 355,000 +/- 15,000 molecules/cell. Using a combination of double-label autoradiography and double-chromagen immunohistochemistry, we determined simultaneously the distribution of (a) i.v. injected D3 MAb; (b) coinjected isotype-matched control IgG (BL3); (c) D3 antigen; (d) blood vessels. The previously developed mathematical models aided in the design of these experiments. Double immunochemical staining of the tumors showed antigen-rich patches 100-800 microns across, surrounded by blood vessels. At a low MAb dose (30 micrograms), binding to antigen severely hindered penetration into antigenic patches as small as 200 microns, even at 72 h. Explanation of this finding by a physical barrier was ruled out by the observation that BL3 distributed uniformly in the same patches. At a higher dose (1000 micrograms), the binding site barrier could be partially overcome. The same general principles of micropharmacology may apply to biological ligands other than antibodies, including those secreted by genetically modified cells.