Background: Gastric cancer (GC), particularly unresectable, metastatic, or recurrent GC, has been characterized by unfavorable prognosis. This meta-analysis of clinical randomized phase II trials was conducted to systematically evaluate the efficacy and safety of capecitabine-based versus S-1-based chemotherapy for metastatic or recurrent GC. Methods: We searched PubMed, Embase, Web of Science, and Cochrane Library databases to identify studies eligible for the present analysis. Data were collected from inception to June 20th, 2019. Outcomes included objective response rate (ORR); 6-, 12-, and 18-month progression-free survival (PFS); 1-, 2-, and 3-year overall survival (OS); and adverse events. A meta-analysis was conducted using a random-effects model, and a sensitivity analysis was conducted to examine whether the results of the meta-analysis were robust. Risk ratio (RR) or hazard ratio (HR) with 95% confidence interval (CI) was reported as the main evaluation parameters. Results: Six eligible studies with 561 subjects were included in the present meta-analysis. There was no significant difference between S-1-based and capecitabine-based chemotherapy in ORR (RR =1.17, 95% CI: 0.95-1.44, P=0.13, I-2 =0%); 6-month (HR =0.94, 95% CI: 0.77-1.14, I-2 =0%), 12-month (HR =0.89, 95% CI: 0.61-1.31, I-2 =0%), and 18-month PFS (HR =1.02, 95% CI: 0.55-1.91, I-2 =0%); 1-year (HR =0.99, 95% CI: 0.83-1.18, I-2 =0%), 2-year (HR =0.90, 95% CI: 0.58-1.42, I-2 =0%), and 3-year OS (HR =1.08, 95% CI: 0.50-2.34, I-2 =0%). However, the capecitabine-based chemotherapy had a higher incidence in all grades of hand-foot syndrome (HFS) (RR =3.41, 95% CI: 1.98-5.90, P<0.01, I-2 =39%) and grades 3-4 neutropenia (RR =1.62, 95% CI: 1.05-2.51, P=0.03, I-2 =0%). Conclusions: In terms of efficacy, capecitabine-based chemotherapy and S-1-based chemotherapy had similar short-term outcomes. Regarding safety, we recommend S-1-based chemotherapy for patients with metastatic or recurrent GC prior to capecitabine-based treatment.