Background: Currently, there are no standard perioperative regimens for human epidermal growth factor receptor 2 positive (HER2+) gastric cancer (GC) or gastroesophageal junction (GEJ) cancer. Furthermore, despite advances in neoadjuvant or perioperative chemotherapy, the efficacy of treatment for locally resectable advanced GC or GEJ cancer remains unsatisfactory. To address this need, we evaluated the effectiveness of adding atezolizumab (programmed death ligand-1 inhibitor) to treatment with trastuzumab (a HER2 antibody) + capecitabine and oxaliplatin (XELOX) compared with trastuzumab + XELOX in patients eligible for surgery with HER2+ locally advanced GC or adenocarcinoma of the GEJ. Methods: Patients aged 18–75 years were randomized 1:1 to atezolizumab + trastuzumab + XELOX (Arm A) or trastuzumab + XELOX (Arm B) and received the regimen for three neoadjuvant cycles (3 weeks per cycle) and five adjuvant cycles. Treatment administration was atezolizumab 1200 mg, trastuzumab 6 mg/kg, oxaliplatin 130 mg/m² intravenously on Day 1, and capecitabine 1000 mg/m² orally twice daily on Days 1–14 of each 3-week cycle. The primary endpoint was the pathological complete regression (pCR) rate. Secondary endpoints were the objective response rate (ORR) during neoadjuvant systemic therapy (NAST) and the R0 resection rate. Time-to-event endpoints are not reported because of immaturity. Efficacy endpoints were analyzed in the intention-to-treat population. The clinical cutoff date for the primary analysis was 12 March 2023. Results: Forty-two Asian patients who provided informed consent were screened and randomized to Arm A (n = 21) or Arm B (n = 21); all patients completed NAST. Most patients were male (92.9%), and the median (range) age was 61 (33–72) years in Arm A and 65 (49–72) years in Arm B. The pCR rate was 38.1% (8/21) in Arm A and 14.3% (3/21) in Arm B (treatment difference: 23.8% [90% confidence interval (CI), 1.3%–44.7%]). The pCR rate was significantly better in Arm A than in Arm B (P = 0.079 <0.1; lower limit of the 90% CI >0). Subgroup analysis showed that age <65 years, male sex, and intestinal Lauren classification were associated with a better pCR rate for the atezolizumab inclusive treatment arm (Arm A); however, these results require further confirmation in future studies. No significant difference was detected between the two arms in ORR during NAST or the R0 resection rate. Conclusions: The addition of atezolizumab to trastuzumab + XELOX therapy was effective in patients with HER2+ locally advanced resectable GC or adenocarcinoma of the GEJ. Clinical trial information: NCT04661150.