We compared efficacy and safety of paclitaxel/capecitabine therapy followed by capecitabine for maintenance (PACX) versus cisplatin/capecitabine therapy (XP) in advanced gastric cancer. Multicenter, randomized, phase III trial was conducted in China (December 2009-February 2014). Adults (n = 320) with histologically confirmed, untreated metastatic/unresectable gastric or gastroesophageal junction adenocarcinoma; with >= 1 measureable lesions according to Response Evaluation Criteria in Solid Tumors 1.0 criteria; Karnofsky performance score >= 70 and life expectancy >= 3 months were randomized (1:1) to PACX or XP. PACX group received paclitaxel 80 mg/m(2) intravenous on days 1 and 8; capecitabine 1000 mg/m(2) orally BD on days 1-14, followed by a 7-day rest interval for 4 cycles, followed by maintenance capecitabine at same dosage/schedule until disease progression, unendurable adverse events or death. XP group received cisplatin intravenous 80 mg/m(2) on day 1 and capecitabine at same dosage/schedule as PACX group per cycle for 6 cycles. Median progression-free survival (5.0 versus 5.3 months; hazard ratio [95% CI]: 0.906; 0.706-1.164; p = 0.44) and overall survival (12.5 versus 11.8 months; hazard ratio: 0.878 [0.685-1.125]; p = 0.30) were not significantly different between PACX and XP groups. Objective response rate was significantly higher (43.1 versus 28.8%; p = 0.012) and disease control rate was similar (77.5 versus 72.5%; p = 0.75) in PACX versus XP, respectively. Quality of life was significantly improved in PACX versus XP after three treatment cycles. Many treatment-related adverse events were significantly lesser in PACX than XP. First-line chemotherapy with PACX is effective with milder toxicities in advanced gastric cancer, but could not replace XP.