Background: 10-20% of GC are HER-2 positive. The role of perioperative anti-HER2-directed treatment is yet undefined. Methods: This randomized, open-label phase II-trial investigates the benefit of combining T alone or with P and perioperative CT for GC and EGJC. Between 2015 and 2021,172 of a planned 215 patients (pts) with centrally confirmed, positive HER-2 status and resectable GC or EGJC (UICC TNM stages Ib-III) were included. Recruitment was prematurely terminated due to slow accrual. Pts were randomized in a 1:2:2 ratio to: Arm A (CT alone) (35 pts); Arm B (CT+ T [8mg/kg, followed by 6mg every 3 weeks]) (67 pts); Arm C (CT + T+ P [840mg every 3 weeks]) (70 pts). CT was initially cisplatin (80 mg/m² d1) and capecitabine (2 x 1000 mg/m²/d d1) for 3 cycles before and after surgery. After publication of the FLOT-4 study, the protocol was amended. CT changed to four cycles FLOT (Al-Batran Lancet 2019) with FOLFOX or CAPOX as alternative for pts ineligible for FLOT. In the experimental arms, T and P were continued beyond CT at the same dose for a total of 17 cycles. Major pathological response rate (mpRR) determined by central pathology review was the primary endpoint. The study was designed to have 80% power to detect an increase in mpRR from 25% with CT to 45% with CT+T+P or CT+T with a one-sided alpha of 10%. CT+T+P was first tested versus CT and if positive, CT+T would be tested versus CT. Results: Out of 172 pts randomized, 161 fulfilled all important eligibility criteria and started their allocated treatment (per protocol population). 62.1% of pts had EGJC and 72.0% an intestinal subtype. Main CT regimens were cisplatin+capecitabine (42.2%) and FLOT (46.6%). In Arm A:B:C, 90.9%, 92.2% and 81.3% completed neoadjuvant treatment. Major reason for treatment discontinuation was toxicity (70%). Surgery was performed in 84.8%, 98.4%, 92.2% pts in Arm A:B:C. R0 resection rates were 83.9%, 90.3% and 85.9%. At present, results of central pathology review of mpRR are available for 126 out of 150 operated pts (84.0%). Pts not operated (n=11) were considered as failures for mpRR. MpRR was 23.3%, 37.0%, 26.4% in Arm A:B:C. The increase of 3.1% (80% CI: [-9.5%, 15.7%], one-sided p=0.378) in Arm C vs. A was not statistically significant. The increase in Arm B vs. A was 13.7% (80% CI: [0.7%,26.7%], one-sided p=0.099). MpRR was 33.3%, 53.3% and 37.9% in Arm A:B:C after amending the protocol while, in contrast, it was 8.3%, 16.7% and 12.5% before. Conclusions: The primary endpoint analysis did not meet the pre-specified criteria of efficacy for the combination of CT+T+P. However, CT+T showed interesting response rates, especially with FLOT as CT backbone. Follow-up data including survival is necessary to define the clinical value of this regimen. Clinical trial information: NCT02205047.