Background: The phase 3 KEYNOTE-585 study (NCT03221426) evaluated neoadjuvant/adjuvant pembrolizumab (pembro) or placebo (pbo) + chemotherapy (chemo) followed by adjuvant pembro vs pbo in locally advanced, resectable gastric or gastroesophageal junction (G/GEJ) cancer. A separate cohort evaluated pembro + FLOT (FLOT cohort). We report results from safety and efficacy analyses of the FLOT cohort at third interim analysis. Methods: In the FLOT cohort, patients (pts) with untreated, locally advanced, resectable G/GEJ cancer were randomized 1:1 to neoadjuvant pembro 200 mg IV Q3W (pembro gp) or pbo (pbo gp) Q3W for 3 cycles + FLOT (docetaxel, oxaliplatin, leucovorin, and 5-FU) Q2W for 4 cycles. After surgery, pts received adjuvant pembro or pbo Q3W for 3 cycles + FLOT Q2W for 4 cycles, then adjuvant pembro or pbo Q3W for 11 cycles. Endpoints evaluated included safety, pathCR rate (BICR), EFS (RECIST 1.1, by investigator), and OS in the ITT. Data cut-off at third interim analysis was 09 Feb 2023. Results: A total of 203 pts were randomized (100 pembro + FLOT; 103 pbo + FLOT). Among these, 12 (6%) pts had cT1-T2, 161 (79%) had cT3, and 9 (4%) had cT4. 140 (69%) had N+ disease, and 79 (39%) had GEJ adenocarcinoma at baseline. Median follow-up was 31.6 months (mo) and 31.1 mo, respectively, at IA3. A total of 94 of 99 (95%) pts completed the neoadjuvant phase, 87 of 99 (87%) completed the surgery phase, and 45 of 77 (58%) completed adjuvant treatment. The R0-resection rate was 79% vs 80% in the pembro gp vs pbo gp, respectively. The pathCR rate was 17.0% (95% CI, 10.2-25.8) in the pembro gp and 6.8% (95% CI, 2.8-13.5) in the pbo gp, estimated difference (10.2% [95% CI, 1.3-19.7]). Median EFS was not reached (95% CI, 28.2-NR) in the pembro gp and 30.9 mo (22.8-NR) in the pbo gp (HR 0.79; 95% CI, 0.52-1.22). The 24-mo EFS rates were 66% and 57%, respectively. Median OS was not reached in either group (HR 1.04; 95% CI, 0.66-1.66). The 24-mo OS rates were 72% and 73%, respectively. Grade ≥ 3 drug-related AE rates in all phases combined were 76% and 63% in the pembro vs pbo gp, with serious drug-related AEs in 42% vs 20%, and surgery-related AEs in 20% vs 13%, respectively. A total of 3 (3%) vs 1 (1%) pt in the pembro vs pbo gp died due to a drug-related AE. Conclusions: Neoadjuvant/adjuvant pembro + FLOT was feasible with no new safety concerns. PathCR and EFS favored pembro + FLOT vs pbo + FLOT in pts with untreated, locally advanced resectable G/GEJ cancer. Clinical trial information: NCT03221426.