Ovarian cancer is the leading cause of death from gynecological malignancies among women in the United States. The 5-year survival for the patients with late stage tumors is 20%, compared to 50–90% in early stage tumors. The aim of this study is to use state-of-the-art molecular technologies to better understand the biology of ovarian cancer. We used cDNA microarrays to distinguish the variation in gene expression of approximately 20,000 genes among 10 early stage (stage I/II) and 10 late stage (stage III/IV) ovarian tumors against 5 pooled normal ovarian epithelial cell brushings. Subtracted cDNA libraries of several of these tumors versus normal ovarian epithelial cell brushings were generated to identify additional genes not present on the cDNA microarrays. Calculation of average fold induction (both up and down) revealed no statistically significant increase in the number of genes showing differential expression in the late stage tumors compared to early stage, and the differentially expressed genes in both the early and late stage tumors were very similar. The loss of expression of 20 of 30 top candidate down-regulated genes was confirmed in a panel of both early and late stage tumors (15 each) by semi-quantitative RT-PCR. To complement the gene expression profiles obtained, DNA from 35 ovarian tumors of various stages/grades were used for comparative genomic hybridization (CGH) and loss of heterozygosity (LOH) studies. Gains were commonly observed on chromosomes 1, 8, 17, 19 and 20, whereas losses were mainly observed on chromosomes 4q, 5q, 13q, and 18q. 13q14.1 and 19q13.4 were two regions that showed more loss in early stage than late stage tumors. Through these analyses we are developing a molecular signature for ovarian cancer and identifying important genes involved in early stage ovarian carcinogenesis.