Mismatch repair-deficient (dMMR) endometrial cancer (EC) is an inflamed phenotype with poor outcomes when meeting high-risk criteria and limited treatment options in the adjuvant setting. We report protocol-prespecified subgroup analysis of patients with dMMR tumors from the phase III ENGOT-en11/GOG-3053/KEYNOTE-B21 study (ClinicalTrials.gov identifier: NCT04634877) in newly diagnosed, high-risk EC after surgery with curative intent. Patients were randomly assigned to pembrolizumab 200 mg or placebo (six cycles) plus carboplatin-paclitaxel (four to six cycles) once every 3 weeks, then pembrolizumab 400 mg or placebo once every 6 weeks (six cycles), respectively. MMR status was a stratification factor. Patients received radiotherapy at investigator discretion. Investigator-assessed disease-free survival (DFS) was a primary end point. No formal hypothesis testing was performed for subgroup analysis. In the intention-to-treat population, 141 patients in the pembrolizumab arm and 140 in the placebo arm had dMMR tumors. At this interim analysis, hazard ratio for DFS favored pembrolizumab (0.31 [95% CI, 0.14 to 0.69]); median DFS was not reached in either group. Two-year DFS rates were 92.4% (95% CI, 84.4 to 96.4) and 80.2% (95% CI, 70.8 to 86.9), respectively. No new safety signals occurred. Longer-term follow-up of outcomes will be evaluated at final analysis. Preplanned subgroup analysis on the basis of the study's stratification factors suggests that pembrolizumab plus chemotherapy improves DFS and is clinically relevant for patients with dMMR tumors in the curative-intent setting.