Based upon contigs coding for centromere specific proteins (CENPs) from a pyrosequencing transcriptome of Saccharina japonica, gene-specific primers were designed for the three centromere specific protein gene cloning from S. japonica. Three CENPs genes, named as SjCENH3, SjNuf2 and SjNCD80 were cloned by using the technique of Polymerase Chain Reaction. The open reading frame of the cloned full-length cDNAs were 432, 1365 and 2172 bp in length, encoding putative proteins composed of 143, 454 and 723 amino acids, respectively. Homologous sequence alignments show: SjCENH3 include the diversity of the N-terminal region and C-terminal histone fold domain (HFD), and HFD including four αhelixes (αN helix, α1 helix, α2 helix and α3 helix) and two loops (Loop1 and Loop2). But compared to H3, SjCENH3 amino terminus is slightly longer and there is no similarity between them, while the sequence of histone fold domain α1-helix, α2-helix and Loop1 are more variable. The main domains of SjNuf2 are Nuf2 domain, Spc7 and SPT2. The main domains of SjNdc80 are SPT2, Spc7, Ndc80_Hec, FoP_duplication and Kin17_mid. Neighbor-Joining (NJ) phylogenetic tree inferred from the putative proteins of CENH3, Nuf2 and NCD80 genes of S. japonica and other species indicated that SjCENH3 and SjH3 genes were divided two clades: CENH3 and H3 clade. H3 is highly conserved across species, while CENH3 differences among species are relatively large. Nuf2 and Ndc80 from different organisms could be significantly divided into two clades, angiosperms and algae. They both may have relations of co-evolution. This study first reported the characteristics of three centromere protein-coding genes SjCENH3, SjNuf2 and SjNdc80, providing a good basis to locate the centromere position by immunofluorescence technique and to get the centromeric DNA sequence by chromatin immunoprecipitation and to analyze chromosome karyotype using preparation of antibodies specific for the corresponding proteins of S. japonica.