SORL1 was shown to be genetically associated with late-onset Alzheimer's disease (LOAD) in a large-scale genome-wide association study (GWAS) involving clinically verified subjects. Here, we attempted to replicate the association of SORL1 in Japanese neuropathologically characterized brain donor subjects (LOAD, 213; control, 370) through a single-nucleotide polymorphism (SNP)-based genetic study involving 19 SNPs: 11 SNPs were selected from the initial study reported by Rogaeva et al. (2007), and the other eight were from our GWAS. Among these SNPs, five exhibited a significant association with LOAD after multiple test correction (p < 2.63E-03 [ = 0.05/19]), which was supported by means of multiple logistic regression analysis with adjustment for age, gender, and carrier status of the APOE ε4 allele. Three of these SNPs (rs985421, rs12364988 [Rogaeva's SNP 7], and rs4598682) were encompassed by a 5' linkage disequilibrium (LD) region, and the remaining two (rs3781834 and rs3781836) by a 3' LD region. Strong LD among SNPs was observed within each LD region, implying that there are two genomic regions showing association with LOAD in SORL1. Case-control haplotype analysis demonstrated that some haplotypes are associated with LOAD in both LD regions. Our replication study strongly supports the preceding evidence that SORL1 is likely one of the genes associated with LOAD.