We used dextran sulfate (DS) to evaluate barrier charge selectivity in 11 nonproteinuric subjects and in 11 patients with the nephrotic syndrome due to either membranous nephropathy or minimal change nephropathy. The 3H-DS preparation spanned a molecular radius interval of 10-24 A and exhibited size-dependent protein binding in vitro. Urine and ultrafiltrates of plasma were separated by size into narrow fractions using gel permeation chromatography. The sieving coefficient (theta) for ultrafilterable DS of 15A radius averaged 0.68 +/- 0.03 in nonproteinuric vs. 0.95 +/- 0.05 in nephrotic subjects (P < 0.001). Uncharged dextrans of broad size distribution were used to evaluate barrier size-selectivity in separate groups of nonproteinuric subjects (n = 19) and nephrotic patients with either minimal change (n = 20) or membranous nephropathy (n = 27). The value of theta for an uncharged dextran of similarly small radius (approximately 18 A) was significantly larger than that observed for DS in nonproteinuric subjects, but was similar in nephrotic individuals. Further, impaired barrier size-selectivity, as assessed by the sieving profile for uncharged dextrans (18-60 A radius), failed to account fully for the observed level of albuminuria in almost half of the patients with either minimal change (9/20) or membranous nephropathy (12/27). Together these findings suggest that the human glomerular capillary wall normally provides an electrostatic barrier to filtration of negatively charged macromolecules such as albumin, and that impairment of this electrostatic barrier contributes to the magnitude of albuminuria in the nephrotic syndrome.