OBJECTIVES: Despite improved knowledge regarding the etiology of ovarian cancer, as well as application of aggressive surgery and chemotherapy, there has been only a modest change in the mortality statistics over the last 30 years. Given these results and the evolution of targeted therapies, there is an increasing need for prognostic and predictive factors to stratify patients for individualized care. Many laboratories have also investigated the specific individual biomarkers correlating them with clinicopathologic characteristics. Unfortunately, the vast majorities of these biomarkers have not proved clinically valuable. In this article, we review published genomic signatures including data generated in our laboratory for their relevance. METHODS: Multiple published expression profiling articles were selected for review and discussion. Genomic studies were separated from those with dichotomized survival data and unsupervised analysis to identify discreet subsets of tumors and studies that generated activated pathways. RESULTS: The identification of prognostic and predictive individual biomarkers has been common. Few of these have been validated. Genomic profiles have been obtained that distinguish short- from long-term survivors. The relevance of these studies to the large number of patients within the extremes remains unclear. Unsupervised clustering studies of ovarian cancers have identified potential subsets of tumors that reflect different clinical behavior. These studies will require large numbers of independent samples for validation. Another approach has been to identify genes that correlate with patient survival as a continuous variable. These genes are then placed into biologic context using pathway analysis. These pathways provide potential therapeutic targets, and those patients whose tumors express these targets may be most effectively treated by using inhibitors specific for the pathway. CONCLUSIONS: There is a major need for prognostic and predictive biomarkers for ovarian cancer. With the development of new genomic technologies, there is an opportunity to identify gene expression signatures that can be used to stratify patients according to their ultimate survival and response to chemotherapy. Large independent sets and robust statistical techniques will be required to fully exploit this approach.