Although many randomized trials of chemotherapy for metastatic gastric cancer have been reported during the past two decades, no standard regimens worldwide have been established yet. Reference arms vary depending on the region and cultural differences. To date, a combination of 5-fluorouracil (5-FU) and cisplatin is most widely used. However, no confirmation of survival advantage over single-agent 5-FU in a randomized trial has been proved yet, and there remain limitations of efficacy results in older-generation regimens. Recently developed new agents such as irinotecan, taxanes (paclitaxel and docetaxel), and new oral fluorouracil (S-1 and capecitabine) provided more promising results: a response rate over 50% and median survival time (MST) over 10 months in their preliminary combination studies. These newer combination regimens are now being investigated in various randomized phase III studies, which will clarify whether the newer-generation regimens provide survival advantage over older-generation regimens. The MST of the new standard should exceed 11 months to be considered a definite improvement, and overall survival seems to be a more desirable primary end point than progression-free survival in a randomized trial. Molecular targeting agents are another concern to improve the treatment outcomes of this disease and are now under investigation in combination with conventional cytotoxic agents. Both clinical and biological research will be more important in future studies.