Nucleotide variation in the alcohol dehydrogenase (Adh1) locus region of the wild rice Oryza rufipogon and its related species was analysed to clarify the maintenance mechanism of DNA variation in Oryza species. The estimated nucleotide diversity in the Adh1 locus region of O. rufipogon was 0.002, which was one of the lowest values detected in nuclear loci of plant species investigated so far. Tests of neutrality detected significantly negative deviation from the neutral mutation model for the coding region, especially for replacement sites. When each of the ADH1 domains was considered, significance was detected only for the catalytic domain 1. These results suggest purifying selection in the Adh1 coding region. In the phylogenetic tree of Oryza species based on Adh1 variation, cultivated rice O. sativa subspp. japonica and indica were included in the cluster of O. rufipogon. The genetic distance of the Adh1 region between O. rufipogon and O. sativa was as low as the nucleotide diversity of O. rufipogon. These results imply that O. rufipogon and O. sativa cannot be classified based on the nucleotide variation of Adh1. No replacement divergence between O. rufipogon and the other three A-genome species (O. glumaepatula, O. barthii and O. meridionalis) were detected, indicating that ADH1 is conserved in the A-genome species. On the other hand, between O. rufipogon and the E-genome species O. australiensis, replacement changes were detected only in the catalytic domain 1. The difference in replacement substitutions between the A- and E-genome species may be related to adaptive changes in the ADH1 domains, reflecting environmental differences where the species encounter anaerobic stress.