Biochemical and molecular genetic studies have contributed to our molecular knowledge of blood group-associated molecules in the past few years. Among the 23 blood group systems presently identified, almost all have a molecular basis and present investigations are oriented towards the analysis of genetic polymorphisms, tissue-specific expression and structure-function relationships. Antigens defined by carbohydrate structures, among which ABO, Hh, Lewis and Secretor are the main representative species, are indirect gene products. They are synthesized by Golgi-resident glycosyltransferases, which are the direct products of the blood group genes. Many of these enzymes have been cloned and the molecular basis of the silent phenotypes, for instance 0, Bombay/paraBombay, Le(a-b-) and non-secretor, has been elucidated. However, the glycosyltransferases involved in the biosynthesis of Pk, P and P1 antigens are not yet characterized. A large number of blood group antigens carried by red cell polypeptides expressed at the cell surface are not related to a carbohydrate structure, and these proteins are direct blood group gene products. Most have been cloned and characterized recently, for instance MN antigens (glycophorin A), Ss antigens (glycophorin B), Gerbich antigens (glycophorins C and D) and antigens encoded by the RH, LW, KEL, FY, JK, XG, LU and XK loci. Other antigens have been located on proteins already identified, for instance the Cromer antigens on DAF, Knops antigens on CR1, Indian and AnWj antigens on CD44, Yt antigens on AChE, Diego, Wr, Rga and Warr on Band 3, Colton antigens on AQP-1 (water channel). The SC (Scianna) et DO (Dombrock) systems, however, still resist to molecular cloning. On the basis of this information, a tentative classification of blood group antigens into five functional categories is emerging: - Transporters and channels, - Receptors and ligands, - Adhesion molecules, - Enzymes, - Structural proteins. This review will focus on these recent findings and will illustrate how these studies may bring new information for analysis of normal and abnormal phenotypes and for understanding both the mechanisms of tissue specific expression and the potential function of these antigens, particularly those expressed in non-erythroid lineage. In addition, since our knowledge of the molecular basis of blood group polymorphisms has significantly increased, new genotyping techniques potentially useful in clinical applications will become available.