The ability of native and chemically modified bovine serum albumin (BSA) to maintain normal pulmonary microvascular permeability was tested in "bloodless," fluorocarbon emulsion exchange transfused rats. Wet-to-dry weight ratios (W/D) of whole lung and morphometric estimates of the amount of ferritin transported to basement membrane were used to assess changes in water flux and macromolecular transport, respectively. Native and modified BSA in capillary walls were localized by immunogold techniques. Arginine residues of BSA were blocked with cyclohexanedione (CHD-BSA), and lysine residues were modified either by succinylation (Succ-BSA) or reductive methylation. Succinylation and CHD modification of BSA caused alterations in antigenicity and trypsin sensitivity; succinylation reduced the isoelectric point (pI). Whereas administration of either CHD-BSA or Succ-BSA increased the W/D, transport of ferritin to basement membrane was greater in the presence of Succ-BSA than CHD-BSA. By contrast, infusion of reductively methylated BSA in which modified lysines altered neither antigenicity nor pI, resulted in a W/D and amount of ferritin in basement membranes comparable to that of BSA. Binding of CHD-BSA and Succ-BSA to endothelial glycocalyx appeared to be reduced relative to native BSA and reductively methlyated BSA. The lowered pI of Succ-BSA may have contributed to its reduced binding; reductively methylated BSA with an unaltered pI was present in the glycocalyx. These data are consistent with a role for positively charged arginine residues in the interaction of albumin with the glycocalyx. The W/D of animals perfused with BSA was higher than those reported for rats perfused with complete rat serum proteins. This is consistent with the notion that serum factors, in addition to albumin, are required to maintain normal microvascular permeability.