BACKGROUND AND OBJECTIVES Reagent red blood cells (RBCs) for antibody detection should express certain important antigens as a double dose, that is, the donors must be homozygous for the corresponding alleles. Traditionally, dose is determined by serological typing and known allele frequencies. However, RHD zygosity cannot be predicted serologically owing to the absence of an antithetical antigen, and FY zygosity is confounded by two variant haplotypes, FY*0 and FY*X. Furthermore, lack of reagents hampers our ability to type for some clinically important antigen pairs such as Do(a)/Do(b). MATERIALS AND METHODS Genomic DNA was isolated from reagent RBC samples. Established, validated methods were used to determine the RHD, FY, and DO genotypes. RESULTS Three of 52 D+ samples gave results that differed from the predicted genotype: two presumed R(1)R(1) samples and an R(2)R(2) sample were shown to be R(1)r' and R(2)r'', respectively. Five of 59 samples that were from presumed homozygotes for either FY*A or FY*B were heterozygous, together with either FY*X (three samples) or FY*0 (two samples). Seventy-five samples tested for DO were DO*A/A (n = 14), DO*A/B (n = 39), or DO*B/B (n = 22). CONCLUSIONS The results show that serologically determined RhD and Duffy phenotypes of reagent RBCs are unreliable and that antigens we thought were represented as a double dose were single dose. The addition of Dombrock genotyping provides information which is useful in antibody identification. We conclude that selected genotype analyses are a valuable quality assurance measure to ensure that reagent RBCs comply with national and international recommendations for test sensitivity.