The production of induced pluripotent stem (iPS) cell lines from human somatic cells, such as skin fibroblasts, recently reported in your January issue1 and elsewhere2–4, opens the exciting new possibility of producing personalized pluripotent stem cell lines from individual patients without using human oocytes or embryos. In addition to circumventing ethical problems associated with human embryonic stem (ES) cells, this new approach could also address immunological rejection associated with cell therapies because iPS cell lines could be created with human leukocyte antigen (HLA)-haplotypes matching those of individual patients. Although such personalized stem cells are possible, the time and cost necessary for the production of clinicalgrade iPS cell lines, the production of differentiated cell types for transplantation and the safety validation of all these procedures (including the final cell products for therapy) could limit wide adoption in clinical practice. An alternative approach for reducing or avoiding immunological rejection of cell therapies would be to construct an HLA-haplotype bank of pluripotent stem cell lines, the possibility of which was first proposed for the UK population5. In a previous study, we calculated the appropriate size of human ES cell lines necessary for providing the majority of the Japanese population with beneficial HLA-matching6. This showed that 200 ES cell lines derived from randomly donated embryos can provide beneficial matching for 80% of people with at least two-locus matching for HLA-A, HLA-B and HLA-DR. Moreover, 100 parthenogenetic ES cell lines from randomly donated oocytes, can be matched to three locuses in 90% of the population because such ES cell lines are homozygotes for all loci. In this context, the recent production of parthenogenetic human ES cell lines is of note7, although clinical application would likely be compromised by abnormal (maternal) genomic imprinting across all loci. If iPS cell lines could be derived from donors’ somatic cells with an efficiency and safety profile that was comparable to ES cell lines, an opportunity exists to establish more efficient HLA-haplotyped iPS cell banks. Under such a scheme, individuals with certain HLA-haplotypes could be chosen for production of iPS cell lines, rather than depending on randomly donated embryos or oocytes. It would be particularly useful if somatic cells, such as a small piece of skin, used to derive iPS cells, could be obtained from HLA-homozygous individuals. Two questions directly pertain to the feasibility of such an approach: first, how many such cell lines would be needed for HLA matching to the majority of iPS cell recipients, and second, how easy would it be to find homozygote donors? To answer these questions, we estimated the frequencies of HLA homozygotes in the Japanese population from the observed frequencies of HLA threelocus haplotypes (AB-DR) at broad serology-level resolution, based on the same data used in our previous study5. This data set comprises HLA-A, HLA-B, and HLA-DR types from a total of 2,578 unrelated individuals, including unrelated