A randomized phase II clinical trial is being conducted for patients with advanced or recurrent gastric cancer, in order to select the most promising treatment for subsequent evaluation in a large-scale phase III trial. We compare four chemotherapeutic treatments, which include two sequential and two combination regimens using paclitaxel with 5-fluorouracil or S-1, an oral fluorouracil derivative. The primary endpoint is 10-month overall survival rate, while the secondary endpoints are adverse events, time to treatment failure and progression-free survival. A Bayesian method is used to provide a statistical rule for monitoring the trial. Forty patients per treatment regimen (160 in total) were randomized into one of the four regimens using a centralized dynamic method.