Polydisperse [3h] dextran was infused into eight Munich-Wistar rats in the early autologous phase of nephrotoxic serum nephritis (NSN), thereby permitting direct measurements of pressures and flows in surface glomeruli and fractional clearances for dextrans [(U/P) dextran/(U/P) inulin] ranging in radius from 18 to 42 A. Despite glomerular injury, evidenced morphologically and by a marked reduction in the glomerular capillary ultrafiltration coefficient, the glomerular filtration rate remained normal because of a compensating increase in the mean net ultrafiltration pressure. In NSN rats, as in normal controls, inulin was found to permeate the glomerular capillary wall without measurable restriction, and dextrans were shown to be neither secreted nor reabsorbed. For dextran radii of 18, 22, 26, 30, 34, 38, and 42 A, (U/P) dextran/(U/P) inulin in NSN and control rats, respectively, averaged 0.90 vs. 0.99, 0.81 vs. 0.97, 0.63 vs. 0.83, 0.38 vs 0.55, 0.20 vs. 0.30, 0.08 vs. 0.11, and 0.02 vs. 0.03. Using a theory based on macromolecular transport through pores, the results indicate that in NSN rats, effective pore radius is the same as in controls, approximately 50 A. In NSN, however, the ratio of total pore surface area to pore length, a measure of the number of pores, is reduced to approximately 1/3 that of control, probably due to a reduction in capillary surface area. These results suggest that proteinuria in glomerular disease is not due simply to increases in effective pore radius or number of pores, as previously believed. Using a second theoretical approach, based on the Kedem-Katchalsky flux equations, dextran permeability across glomerular capillaries was found to be slightly lower, and reflection coefficient slightly higher in NSN than in control rats.