Recently, the reconstruction of co-regulated gene network has become increasingly popular in the area of bioinformatics. It tries to find the associated information and network topology among genes through large numbers of biological data. In this paper, we proposed a novel method PC-GO by using path consistency (PC) algorithm based on gene ontology (GO). GO provides a basis for measuring the conditional semantic similarity between genes, and then PC algorithm is applied to remove links between genes with less correlation in the network. We successfully applied our algorithm to yeast data. Experimental results show that the accuracy and integrity of the co-regulated network acquired by our method outperforms previous methods.