BACKGROUND: The response evaluation criteria in solid tumor (RECIST) exclude the use of barium meal studies. This will deeply affect the Japanese criteria for evaluating the response in the primary lesion of gastric cancer.METHODS: Of 280 patients with gastric cancer enrolled in a Japan Clinical Oncology Group (JCOG) phase III study, 255 had been assessed for response by the WHO and/or Japanese criteria. We selected these 255 patients as our subjects and reassessed their response outcomes by RECIST.RESULTS: Of the 255 patients, 32 (13%) had no evaluable lesion other than the primary site, and 171 (67%) had some measurable lesion defined by the WHO criteria. Because the lesions in 129 of these 171 patients were 20 mm or more in size, only 51% of the 255 subjects were eligible for assessment of the target lesion by RECIST. In 162 of the 171 patients who had a lesion of 10 mm or more, response rates by the old and new criteria were nearly equal, in spite of the different methods of measurement. The response rate in the primary lesions assessed by the Japanese criteria was lowest among all the groups examined.CONCLUSION: The RECIST is simple and good for clinical practice. Nevertheless, the recommendation of 20 mm or more for target lesions, which may restrict the number of eligible patients, could be replaced by a size of 10 mm or more. Because the Japanese evaluation criteria are rigid and do not inflate the response rate, they can be used, as additional criteria for assessing the quality of response, when the RECIST is used.