Leucovorin (LV) enhances the activity of 5-fluorouracil (5FU). Based on these data, we performed a randomized trial with 5FU, epirubicin (EPI), mitomycin C(MMC) with/ without LV in advanced gastric cancer (AGC). The purpose of our study was to investigate if the addition of LV improved the response rate of the combination 5FU EPI, MMC (FEM) over FEM. From January 1988 until April 1994, 88 patients with recurrent or metastatic AGC were randomly received 5FU, EPI, MMC with (group A) or without (group B) LV. Between the two arms of the study no difference was noticed in sex, performance status, primary site of tumor, and lymph node metastases. Therapy included group A (5FU 600 mg/m2/day, i.v. bolus, on days 1, 8, 29, 36, and EPI 45 mg/m2/day, i.v. bolus, on days 1 and 29, MMC 10 mg/m2/day, i.v. bolus, on day 1) and group B (the same as group A plus LV 200 mg/m2/day by 2 h intravenous infusion with 5FU intravenous push at midinfusion). No significant difference in response rate was noticed between the two treatment arms; there were two (5%) patients with complete response in group A, and five (12%) in A and 11 (26%) partial responders in group B (p < 0.1). A significantly higher number of patients achieving stable disease was observed in group B; 19 (44%) in comparison to group A 10 (24%) (p < 0.048). There were more patients with progressive disease in group A 25 (59%) than in group B 12 (28%) (p < 0.003) (Table 2). No difference was noted in mean duration of response: group A, 15.8 (6-31) weeks; and group B, 17.6 (6-28) weeks. The mean time to progression was for group A [11.4 (6-35) weeks] and for group B [17.6 (8-33) weeks]. Mean survival was for group A [27.4 (12-59) weeks] and for group B [30.6 (17-53) weeks], for 50% of patients. Causes of death were, for group A, 40 patients from disease progression and two sudden deaths; for group B, causes of death were for 41 patients disease progression and two sudden deaths. There were two patients in group A and one in group B that were not evaluable because they abandoned therapy after the first cycle. Toxicity was increased in group B; anemia, nausea and vomiting, and alopecia (p < 0.055) were more severe in group B, but not statistically different when compared to group A. Neutropenia, thrombocytopenia, mucositis, and fatigue of any grade were significantly more common and severe in group B. Significant dose reductions due to toxicity were required more commonly in group B. We conclude that the response rate was increased in the schedule with the addition of LV, at the cost of increased toxicity and with no difference in survival. A randomized trial comparing FEM-LV with new generation regimens would determine whether the addition of LV qualifies FAM equally active with these.