Background: The PFS with POF was statistically significantly improved and IP PAC was trending to improve compared to FOLFOX in first-line setting in AGC were reported in 2019 ASCO-GI (abstract 6). Update and subgroup analysis were released herein. Methods: The patients with AGC were randomized to three groups. The POF or IP PAC was paclitaxel 135 mg/m²intravenously (POF) or paclitaxel 80 mg/m² intraperitoneally (IP PAC) followed by mFOLFOX6 omitted 5-Fu bolus. Every 14 days repeated for all three regimens. Up to 9 cycles of treatment were administered, followed by S-1 until disease progression. The primary endpoint was PFS. Results: Between Nov 2015 and May 2018, 89 pts (30 POF, 29 IP PAC, 30 FOLFOX) were randomly allocated. PFS, OS and RR were seen in the table below. Either POF or IP PAC was statistically significantly better than FOLFOX in PFS. In subgroup with female, peritoneal metastasis, ascites, lymphadenopathy in peritoneal cavity, number of organs involved > 2, POF was statistically significantly better than FOLFOX in PFS. In subgroup with female, gastrium of primary tumor site, peritoneal metastasis, ascites, no lymphadenopathy out of peritoneal cavity, IP PAC was statistically significantly better than FOLFOX in PFS. Intravenously docetaxel plus S-1 still saw response after IP PAC. Conclusions: either POF or IP PAC improved survival compared to FOLFOX, especially in patients with female or peritoneum metastasis. Only POF, not IP PAC, improved response rate compared to FOLFOX. Clinical trial information: NCT02845908