Epithelial ovarian cancer kills almost 16 000 women each year in part due to late stage of presentation and lack of reliable biomarkers for disease detection. CA-125, the currently accepted serum marker, alone lacks the sensitivity for early stage diagnosis, as only 50% of early stage cases are detected with this marker. Although more early stage cases may be detected by lysophosphatidic acid, this marker is also elevated in other cancers. One major objective of the NCI-FDA Tissue Proteomics Initiative has been to combine the technique of laser capture microdissection (LCM) of epithelial tumor cells in human tissue specimens with two-dimensional gel electrophoresis (2-D PAGE) to identify proteins that may serve as invasive ovarian cancer-specific biomarkers for early detection and/or new therapeutic targets. We performed 2-D PAGE on lysates from five microdissected ovarian tumors (three invasive ovarian cancers and two noninvasive, low malignant potential (LMP) ovarian tumors). We then compared silver stained 2-D gels created from microdissected lysates with SYPRO-Ruby stained 2-D PAGE profiles of the patient-matched undissected bulk tumor lysates from all five patients. Twenty-three proteins were consistently differentially expressed between both the LMP and three invasive ovarian tumors in the limited study set. Thirteen were uniquely present in all three of the invasive ovarian cancer cases and absent or underexpressed in the two LMP cases. Ten were uniquely present in the LMP cases but absent or underexpressed in all invasive ovarian cancer cases. Credentialing and preliminary target validation of the mass spectrometry identified proteins cut from the Ruby-red stained gels was performed by LCM coupled Western blot and reverse-phase array technology in a study set of six cases (the aforementioned five cases used in the 2-D PAGE profiling component of the study plus one additional LMP case). The analysis revealed that the 52 kDa FK506 binding protein, Rho G-protein dissociation inhibitor (RhoGDI), and glyoxalase I are found to be uniquely overexpressed in invasive human ovarian cancer when compared to the LMP form of this cancer. The direct comparison of LCM generated proteomic profiles of invasive vs. LMP ovarian cancer may more directly generate important markers for early detection and/or therapeutic targets unique to the invasive phenotype.