Background: Malignant peritoneal mesothelioma (MPeM) is a rare cancer (incidence of 0.12 per 100,000) and an arduous diagnostic challenge, requiring skillful evaluation by expert pathologists. Subjective and experiential immunophenotyping coupled with inattention to this rare entity causes delayed or misdiagnosis and leads to suboptimal treatment and outcomes. Gene-expression-based cancer classification offers an objective molecular assessment of the tissue of origin for cancers with uncertain diagnosis and improves diagnostic accuracy for MPeM compared to standard pathology. Methods: We retrospectively evaluated 23 patients (pts) with cancer of unknown primary (CUP) or uncertain diagnoses, with peritoneal/retroperitoneal carcinomatosis and reported as malignant mesothelioma using a 92-gene assay (CancerTYPE ID), a validated classifier for predicting tumor types, between January 2013 to December 2016. Pathology was verified by central review. Clinicopathologic data including immunohistochemistry (IHC) was collected using archived pathology specimens and reports. Results: Pts had a median age of 64 years (range: 39 – 93) at diagnosis, 57% were females and 74% had biopsy from omentum/peritoneum. Original histopathology was reported as poorly differentiated in 81% and as adenocarcinoma, carcinoma and malignant neoplasm in 10 (43%), 7 (30%) and 6 (27%) of cases, respectively. The number of IHC stains performed ranged between 0 and 28 (median: 10). Key IHC stains are shown in the table. Conclusions: In this cohort of CUP or uncertain diagnoses with a molecular cancer classification of malignant mesothelioma, we found huge variability and critical gaps in pathological work-up. Integration of molecular cancer classification using the 92-gene assay helped resolve this diagnostic uncertainty, further supporting its clinical utility to complement standard pathology. The impact of these findings goes beyond just MPeM and may benefit numerous pts with orphan cancers, who account for nearly 20% of all cancers diagnosed globally.