This review discusses recent advances in our understanding of the structure, function and molecular genetics of the membrane domain of red cell anion exchanger, band 3 (AE1), and its role in red cell and kidney disease. A new model for the topology of band 3 has been proposed, which suggests the membrane domain has 12 membrane spans, rather than the 14 membrane spans of earlier models. The major difference between the models is in the topology of the region on the C-terminal side of membrane spans 1-7. Two dimensional crystals of the deglycosylated membrane domain of band 3 have yielded two and three dimensional projection maps of the membrane domain dimer at low resolution. The human band 3 gene has been completely sequenced and this has facilitated the study of natural band 3 mutations and their involvement in disease. About 20% of hereditary spherocytosis cases arise from heterozygosity for band 3 mutations, and result in the absence or decrease of the mutant protein in the red cell membrane. Several other natural band 3 mutations are known that appear to be clinically benign, but alter red cell phenotype or are associated with altered red cell blood group antigens. These include the mutant band 3 present in Southeast Asian ovalocytosis, a condition which provides protection against cerebral malaria in children. Familial distal renal tubular acidosis, a condition associated with kidney stones, has been shown to result from a novel group of band 3 mutations. The total absence of band 3 has been described in animals-occurring naturally in cattle and after targeted disruption in mice. Some of these severely anaemic animals survive, so band 3 is not strictly essential for life. Although the band 3-negative red cells were very unstable, they contained a normally-assembled red cell skeleton, suggesting that the bilayer of the normal red cell membrane is stabilized by band 3 interactions with membrane lipids, rather than by interactions with the spectrin skeleton.