In this paper, we present evidence that there are two closely linked copies of the ATP3 gene coding for the γ subunit of the F 1 F 0 -ATPase complex (EC3.6.1.34) in four laboratory strains of Saccharomyces cerevisiae, even though the yeast genome project has reported that ATP3 is a single-copy gene on chromosome II. We previously reported that the gene dosage (three copies) of ATP1 and ATP2 is coincident with the subunit number of F 1 -α and F 1 -β, but that the gene dosage of ATP3 was not consistent with the subunit stoichiometry of F 1 F 0 -ATPase. By applying long PCR and gene walking analyses, we estimated that the two copies of ATP3 were approximately 20 kb apart, and we designated that which is proximal to the centromere ATP3a, while we named that which is distal ATP3b. The nucleotide sequences of the two copies of ATP3 were identical to the reported sequence in the W303-1A, W303-1B and LL20 strains, while only the DC5 strain had a single base substitution in its ATP3a. With the exception of this substitution, the other nucleotide sequences were identical to the upstream 860 bp and the downstream 150 bp. The differences between ATP3 with the single base substitution (Ser 308 to Phe) and ATP3 without the substitution on the complementation of the ATP3 disruptant and on the maintenance of the mitochondrial DNA were observed, suggesting that Atp3ap and Atp3bp in the DC5 strain might have different functions. However, it should not always be necessary for yeast cells to carry different types of ATP3 because the other three strains carry the same type of ATP3. It was also demonstrated that the disruption of the ATP3 genes basically leads to a loss of wild-type mtDNA, but the stability of the mtDNA is not dependent on the ATP3 alone.