Mitochondria are a vital organelle of eukaryotic cells. Many diseases are caused by the mitochondrial gene mutation , therefore, there is an urgent requirement to explore the proteins of mitochondrion. In this paper we have created a list of 1086 human mitochondrial genes by combining 13 datasets from publications or genomic experiments. Then we created a functional linkage network among mitochondrial genes by integrating multiple lines of biological evidence, and utilized the network to predict the function of the mitochondrial genes, including identifying some novel components involved in the KEGG (Kyoto Encyclopedia of Genes and Genomes) pathways and the MIPS complexes. Our results indicated that it is very useful to study mitochondrial biology and disease by integrated genomic and proteomic datasets.