The discovery of Rh partial D variant red cells by discrepant reactions with different monoclonal anti-D has demonstrated the range of Rh D epitopes that have arisen due to alterations in Rh D protein structure. There are two current classification systems, one which uses a nine epitope model (epD1-epD9) whereas a more recent model proposes 30 different epitopes. We describe here the molecular basis of two D variants which lack epD4 and epD9 namely the DNU and D(II) phenotypes. These would have both been originally classified as D(II) phenotype individuals, but we have revealed subtle differences in the serological profile of these erythrocytes. Such a differential reactivity and determination of the molecular bases of these phenotypes allows us to predict critical amino acids for epD3, epD4 and epD9 expression. The DNU phenotype arises from a single point mutation in the RHD gene resulting in a single amino acid change (Gly353Arg). Sequence analysis of exon 7 of the RHD gene derived from the D(II) propositus indicates that there is a single point mutation in this exon resulting in a single amino acid change (Ala354Asp). It is likely that this point mutation gives rise to the D(II) phenotype. Both mutations result in the change to Rh D-specific residues. Our results indicate that the following amino acids are crucial for epD3a (Asp350), epD3b (Asp350 + Gly353), epD4a (Gly353 + Ala354), epD4b (Ala354), epD9a (Asp350 + Gly353 + Ala354) and epD9b (Asp350 + Ala354) expression. All of these amino acids reside on the predicted sixth external domain of the Rh D protein, so it is possible that epD3, 4 and 9 are continuous epitopes.