We found that in our laboratory the MLC was interpretable in most cases, in spite of patient disease state and shipping blood from unrelated donors over great distances. Our data clearly show that DRB1 allele typing was a good predictor of MLC reactivity and we believe that MLC testing in DRB1 allele mismatched unrelated pairs is not informative. However, our data suggest that MLC testing in DRB1 allele matched unrelated pairs is informative, because many such pairs produced weak but significant allostimulation. These weak MLC reactions may guide improvements in matching unrelated donors for BMT. For example, our MLC data suggest that it is important to match unrelated donors at DRB3 alleles for BMT.