Abstract Purpose: The purpose of this study was to evaluate the performance of a proposed prognostic molecular classification of endometrial cancer in a retrospective cohort of n=604 patients. A secondary goal was to compare the performance to existing prognostic models. Methods: All cases were analysed using 1) sequencing of the exonuclease domain of the POLE gene, 2) microsatellite instability by immunohistochemistry, and 3) TP53 mutation, as evaluated by immunohistochemistry. Additionally, complete histopathological and clinical data was available. The cases were classified using the molecular markers analyzed. Results: The assembled cohort largely resembled the typical unselected distribution of FIGO grade and stage. Analysis of clinicopathologic variables showed that FIGO grade and stage, LVSI, tumor histotype, depth of myometrial invasion, ESMO 2013 and 2016 risk group, recurrence and survival all were significantly associated with molecular subgroups. A large number of patients showed multiple molecular aberrations (so called multiple classifiers, n=50, 8.4%). Mutation in P53 was a statistically prognostic marker for overall survival (OS) in the univariable Cox regression (p-value 0.003 for OS, and <0.001 for DSS and PFS, HRs of 1.61, 1.83 and 1.72). The POLE mutated cases showed a trend towards favorable prognosis. By contrast, histotype (endometrioid vs. non-endometrioid) was a highly significant predictor of OS, DSS and PFS (HR=2.23, p=<0.001). Both ESMO 2013 and 2016, which represent an aggregate of the various clinicopathologic factors, were highly significant in predicting outcome and especially so for the high and advanced risk groups with HRs up to 70.1 (CI 23.9-206.8) and all p-values of <0.001. In the multivariable Cox regression analysis, however, none of the molecular markers remained significant. The Harrell C-index for 2016 ESMO risk groups was 0.76. Adding molecular markers did not show improved discriminatory ability. Conclusions: In this retrospective cohort, molecular markers provided important biological information that may be helpful in therapeutic decision making. However, the molecular markers did not outperform current prognostic models. Citation Format: Denis Nastic, Sara Imboden, Mehran Ghaderi, Filippa Rydberg, Daniel Olsson, Michael Mueller, Tilman T. Rau, Elisabeth Epstein, Joseph Carlson. Molecular classification of endometrial cancer provides complementary information but does not outperform current predictive models: The Karolinska and Bern experience [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 465.