Band 3 protein has been incorporated into lipid vesicles consisting of 94:6 (molar ratio) egg phosphatidylcholine-bovine heart phosphatidylserine or total erythrocyte lipids by means of a Triton X-100 Bio-Beads method, with an additional sonication step prior to the removal of the detergent. This method results, for both types of band 3 lipid vesicles, in rather homogeneous vesicles with comparable protein content and vesicle trap. Freeze-fracture electron microscopy revealed that band 3-egg phosphatidylcholine-bovine heart phosphatidylserine vesicles have considerably more intramembrane particles as compared to the band 3-erythrocyte lipid vesicles. The dimensions of the nonspecific permeation pathways present in the band 3-lipid vesicles were measured using an influx assay procedure for nonelectrolytes of different size, in which the vesicles were sampled and subsequently freed from nonenclosed labeled permeant by means of gel-filtration. The band 3-egg phosphatidylcholine-bovine heart phosphatidylserine vesicles have nonspecific permeation pathways (pores), with diameters of up to 60 ~,. In contrast, the band 3-total erythrocyte lipid vesicles are more homogeneous and show much smaller nonspecific permeation pathways, having a diameter of about 12 ji,. These results suggest that the nonspecific permeability of the band 3-lipid vesicles is strongly lipid-dependent. Increase in specific anion permeability expected as a consequence of the presence of band 3 in the erythrocyte lipid vesicles was found to be very limited. However, stereospecific, phioretin-inhibitable D-glucose permeability could clearly be demonstrated in these vesicles. The difference of the nonspecific permeability of the band 3-egg phosphatidylcholine-bovine heart phosphatidylserine vesicles and band 3-erythrocyte lipid vesicles, is discussed in the light of the presence of defects at the lipid/protein interface and protein aggregation, which may induce formation of pores.