3519 Background: AJCC v.7 staging and clpath covariates do not accurately predict outcome of CC patients (pts) eligible for adjuvant chemotherapy (adjCh). Mol markers, microsatellite instability (MSI) and mutations (mut) in BRAF and KRAS, may further improve prognostication in high-risk CC. Methods: Data from 4,796 stage II/III pts accrued to phase 3 trials PETACC3 and N0147 (train set) were used to construct Cox models for overall survival (OS). Variables included stage (pT, pN); clpath (# lymph nodes [LNs] positive, # LNs assessed, tumor grade, primary site, age, gender, adjCh); and mol (MSI, BRAF and KRAS mut). Final models (1. TNM; 2. TNM + mol; 3. TNM + clpath; 4. TNM + clpath + mol) were internally (train set) and externally (test set) validated on 597 stage II/III treated pts from non-clinical trial cohorts. Results: All examined traits were statistically significant for OS prediction (Table 1). We found significant interaction between site and MSI (p=.03) in multivariate models, with worse OS for MSI vs MSS left CC (HR 1.22) and better OS for right CC (HR 0.70). Models 1 to 4 associated with C-indices of 0.65, 0.68, 0.70, 0.72, respectively, in train set; and 0.68, 0.71, 0.73, 0.74 in test set. Corresponding time-dependent AUCs (5 years summary) were 0.54, 0.66, 0.73, 0.74 in train set; and 0.55, 0.68, 0.72, 0.73 in test set. Conclusions: Incorporation of mol markers (MSI, BRAF and KRAS mut) improves prognostic estimation in stage II/III CC pts treated with adjCh. We propose their prospective assessment in larger clinical cohorts, including untreated pts, and development of online calculators that could aid in prognostication (model-based staging) and patient/physician communication.