There is still a lack of satisfactory tumor markers for gastric cancer (GasC). This study is aimed at optimizing parameters in CE-MS based on moving reaction boundary (MRB) so as to improve its sensitivity and stability, and at searching for potential tumor markers of GasC in patients' urine samples via MRB-CE-MS. In this study, several parameters of MRB-CE-MS were investigated and optimized in order to gain optimal stability, sensitivity, and specificity, and was afterwards evaluated as valid. Subsequently, urine samples from GasC patients and control subjects were subjected to MRB-CE-MS analysis under optimized conditions, which successfully distinguished GasC patients from controls, as well as early-stage patients from advanced stage patients. Differentiation performance was evaluated by area under the curve, which showed fine differential value between GasC patients and controls, as well as between early and advanced stage patients (area under the curve value 1.0 and 0.847, respectively). In conclusion, this study established a set of feasible and useful methodology in searching potential tumor markers in urine samples from GasC patients. Moreover, several amino acids have been recognized as potential tumor markers that deserve further investigation.