To investigate the optimal dosage schedule for 5'-deoxy-5-fluorouridine (5'-DFUR), a randomized comparative study was performed in patients with inoperable or advanced gastric cancer. Eight hundred mg/m2/day of 5'-DFUR was administered in the continuous administration group, while in the intermittent administration group 1,400 mg/m2/day was given for 4 days followed by 3 days with no medication in a one-week cycle, making the total dose 5,600 mg/m2/week in both regimens. In both groups, 5'-DFUR was administered for more than 4 weeks. Overall response rate was 14.3% (3/21) for both administration methods. The response rate for the primary site was 9.5% (2/21) in the continuous administration group, and 14.3% (3/21) in the intermittent group. As to toxicity, the incidence of diarrhea was 26.7% in the continuous group and 4.2% in the intermittent group. Although there was no significant difference between these two dosage schedules in efficacy and toxicity, the lower incidence of diarrhea suggested the clinical usefulness of the intermittent administration method. This result emphasized the need to further studies to investigate the most appropriate dosage schedule for fluorinated pyrimidine derivatives.