In the past decades, many high-throughput studies have been performed to investigate molecular mechanisms underlying epithelial ovarian cancer (EOC), to improve treatments and to develop early detection and staging biomarkers. EOC is still a deadly disease due in part to a lack of screening tools and to the absence of subtype and stage-specific targeted treatments. Here, we applied an integrative three-dimensional clustering algorithm to analyze gene expression data from normal ovaries and four subtypes of EOC. Our analysis revealed major differences between subtypes and highlighted biological patterns linked with stages of the disease. These results may contribute to the understanding of molecular mechanisms underlying EOC and find applications in EOC detection and treatment.