The molecular characteristics of recurrent ovarian cancers following chemotherapy treatment have been poorly characterized. Such knowledge could impact salvage therapy selection. Since 2008, we have profiled 168 patients' ovarian cancers to determine the expression of proteins that may predict chemotherapy response or are targets for drugs that are in clinical trials for ovarian cancer treatment. Expression of epidermal growth factor receptor (EGFR), HER2, VEGF, ER, c-Met, IGF1R, Ki67, COX2, PGP/MDR1, BCRP, MRP1, excision repair complementation group 1 (ERCC1), MGMT, TS, RRM1, TOPO1, TOP2A, and SPARC was measured by immunohistochemical analyses at Clinical Laboratory Improvement Amendments-certified laboratories. Our univariate analysis of 56 primary and 50 recurrent tumors from patients with advanced stage ovarian serous carcinoma revealed that PGP and ERCC1 were significantly upregulated in recurrent lesions (P < 0.05). To determine whether these or any of the other markers were differentially expressed in specimens obtained from the same individual at diagnosis and at recurrence, we analyzed 43 matched tumor specimens from 19 advanced stage ovarian carcinoma patients. We confirmed the expression differences in PGP and ERCC1 that were observed in the cohort analysis but discovered that the expression levels of BCRP, RRM1, and COX2 were also discordant in more than 40% of the matched tumor specimens. These results may have implications both for the use of biomarkers in therapy selection as well as for their discovery and validation. Expression of these and other candidate response biomarkers must be evaluated in much larger studies and, if confirmed, support the need for profiling of recurrent tumor specimens in future clinical trials.