Current pathway enrichment method is mainly based on the gene that are differentially expressed, and no enrichment method considers pathway variability (variance). We observed that in the phenotype of disease, some pathways have a significant increase or decrease in variability describing appropriate statistics. Therefore, in this article, we hypothesize that the variation of single pathway is significantly different between two phenotypes. We designed fourteen types of statistics coupled with their test methods to analyze pathways variation and the pathways enrichment significance between two phenotypes, and we compared the results with those obtained by document retrieval. At the same time, the results of five different data preprocessing methods on data were investigated. The results show that RMA is stable in the five gene expression data preprocessing methods. The pathway variation is different between the two phenotypes. According to the literature research results, the permutation test coupled with the variance of Euclidean distance of each gene (the eleventh method) can identify significant pathways more efficiently than GSEA. In conclusion, pathway enrichment analysis strategy based on the pathway variation is feasible, which could be a theoretical guideline for enrichment analysis and a new biological insights of study in human diseases.