In order to screen for antigen-negative blood donors, it is necessary to have appropriate, potent antisera in sufficient volume. Anti-Do(a) and anti-Do(b) are notoriously weakly reactive antibodies, available only in small volumes, usually in sera containing other alloantibodies, and often deteriorate on storage. Thus, it has not been possible to test large numbers of blood samples to find Do (a-) or Do (b-) blood donors. At the NYBC, we now type selected donors for DOA and DOB by DNA analysis. Initially, we tested DNA prepared from donors who had been typed by hemagglutination for one or both antigens. We found that four donors, whose RBCs previously typed as Do (a+b-), had both DOA and DOB alleles, and when retested, the RBCs were Do (a+b+w). We have now tested over 300 donors for DO by PCR-RFLP using either Eam1105 I or BseRI restriction enzymes. Blood from DOA/DOA donors has survived better than "crossmatch compatible" blood for patients with anti-Do(b) and such results suggest that anti-Do(b) is a more frequent cause of transfusion reactions than reported. Furthermore, we have demonstrated that PCR-RFLP can be used to screen for antigen-negative donors in other blood group systems when appropriate antisera are not available. When interpreting the results, it is important to remember that the genotype may not reflect the phenotype. Our strategy has been to perform DNA analysis for the DO alleles on those donors who have been shown by hemagglutination to lack antigens corresponding to multiple alloantibodies in patients' plasma. In this way, we have been able to supply rare blood to numerous patients, whose serum contained at least 5 additional alloantibodies of clinical significance.