Enzymatic removal of blood group A and B antigens from the surface of red blood cells to develop universal blood was a pioneering vision originally proposed more than 25 years ago. A great variety of enzymes, potentially suitable for enzymatic conversion of red blood cells, has been described since, but the process has not been economically viable because of the poor kinetic properties and low pH optimum of enzymes. Recently, the identifi cation of two new families of bacterial glycosidases with enhanced kinetic properties for the removal of A and B antigens at neutral pH marked a milestone in the fi eld of transfusion medicine (Liu et al. 2007). Here we present a detailed structural analysis of Elizabethkingia meningosepticum aN -acetylgalactosaminidase (NagA) shown to effi ciently cleave the A antigen. NagA, a member of glycoside hydrolase (GH) family 109, employs an unusual catalytic mechanism involving NAD +. Comparison of the active-center structure with that of members of GH family 4 reveals a striking degree of structural similarity that allows the postulation of a common reaction mechanism and illustrates a beautiful example of convergent evolution.