Background: OS is the gold standard efficacy measure in oncology; however, its need for prolonged follow-up motivates the establishment of SEs for earlier assessments of emerging treatments. We assessed DFS as a candidate SE for OS in early-stage EC/GEJC using data from Medicare beneficiaries with cancer in the SEER registry. Methods: Patients aged > 65 years in the US with resective surgery after a primary diagnosis of stage 2 or 3 EC/GEJC between 2009-2017 were analyzed (N=925; median follow-up 26.2 months). Surrogacy was assessed via both individual-level association of DFS to OS, and via the association between the treatment effects on DFS and OS. Strength of individual-level association was measured via Spearman’s rank correlation (ρ) non-parametrically and Kendall’s τ using copula functions. The strength of correlation between the treatment effects on DFS and OS—measured by the coefficient of determination (R²) and the surrogate threshold effect (STE), which is the minimum DFS benefit that would translate into statistically significant OS benefit—was derived from a regression model predicting OS hazard ratio (HR) from DFS HR. Patients were classified in clusters based on treatments they received and baseline characteristics (age, sex, index year, staging, and race/ethnicity). Propensity score matching addressed imbalances in baseline characteristics between the treatment and control groups in the constructed clusters. Predictive accuracy of the surrogacy equation was assessed internally via leave-one-out cross-validation and externally via predictions made for 26 RCTs of early-stage EC/GEJC. Results: Patients were mostly male (84%), non-Hispanic white (89.3%), with median age of 71.8 years and almost evenly distributed between cancer stages 2 (50.4%) and 3 (49.6%). Among patients receiving adjuvant (23.6%) or neoadjuvant (82.8%) treatment, most (81.7% of adjuvant and 92.0% of neoadjuvant therapies) received multi-agent chemotherapy. Spearman’s ρ was estimated to be 0.76 (95% CI: 0.70, 0.89) whereas estimates for Kendall’s τ ranged between 0.62 and 0.79. Estimated R² for the correlation between treatment effects was 0.92 (95% CI: 0.56, 1.00) and estimated from the surrogacy equation log(HR_OS) = 0.02 + 1.09 × log(HR_DFS) with a corresponding STE of 0.86. The 95% prediction intervals generated from this equation contained the raw OS HRs for 91% of the clusters in the internal validation, and 89% of the RCTs in the external validation. Conclusions: Correlations between DFS and OS, and between the treatment effects on these endpoints, were both moderate. The highly accurate surrogacy equation between the treatment effects can enable earlier assessments of OS benefit from the DFS benefit for early-stage EC/GEJC treatments in the real-world setting.