Graft-vs-host disease (GVHD) is an adverse effect of allogenic bone marrow transplantation. Although a major cause of GVHD following bone marrow transplantation is incompatibility of major histocompatibility antigen (human leukocyte antigen, HLA) in donor-recipient pairs, the incompatibility of minor histocompatibility antigen (mHa) is known as another cause, especially in HLA-matched donor-recipient pairs. In 1998, Lunetta and Rogus proposed the use of discordant sib-pair (DSP) linkage analysis for detecting mHa and calculated the statistical power using the GVHD model, assuming single mHa locus with multiple alleles. Recently, we proposed a different GVHD model, assuming multiple mHa loci with two alleles (biallelic), considering the single-nucleotide polymorphisms. When the effect of each mHa locus on the occurrence of GVHD is independent, the possible triangle for DSP proposed by Lunetta and Rogus is not optimum, but a new possible triangle, named here as GVHD region, is needed. We evaluated, based on Monte Carlo simulation, the test criteria [log of odds (lod) score cutoffs] and power of DSP using the GVHD region for various parameter sets. The GVHD region showed a higher power than the DSP and entire regions in plausible situations. Our results suggest that the application of GVHD region to DSP is effective for the screening of mHa loci.