The technical developments and expanded indications for testicular sperm extraction (TESE) with intracytoplasmic sperm injection (ICSI) provide great advantages for patients with non-obstructive azoospermia. Such success, however, also means that genetic abnormalities in non-obstructive azoospermia can be transmitted to the next generation, demonstrating the importance of being able to understand the genetic background of non-obstructive azoospermia. We have previously reported that human leukocyte antigens (HLA)-A33 and -B44 in the HLA class I region and the HLA-DRB1*1302 allele in the HLA class II region are linked to susceptibility to non-obstructive azoospermia in Japanese men. However, strong linkage of HLA-DRB1*1302 with HLA-A33 and -B44 is also evident in the Japanese population. Thus, uncertainty prevails as to whether the HLA class I or class II molecule is more directly associated with non-obstructive azoospermia. In the present study, we performed association analysis with 21 polymorphic microsatellite markers identified near the HLA genes to map the gene involved in the development of non-obstructive azoospermia more precisely. Microsatellite markers located in the HLA class I region or the class III region showed no statistically significant association with this disorder, although once again the HLA-A33 and -B44 alleles showed a significant association. In contrast, some of the microsatellite markers in the HLA class II region and at the HLA-DRB1 and -DQB1 loci displayed strong associations with non-obstructive azoospermia. Taken together, our previous and present data suggest that the critical region for development of non-obstructive azoospermia is near the HLA-DRB1 and -DQB1 segments in the HLA class II region.