The aim of this trial was to investigate whether pre-operative chemotherapy leads to a 15% higher curative resectability rate in patients with operable gastric cancer. In this randomised trial, patients were allocated to receive either four courses of chemotherapy using 5-fluorouracil, doxorubicin and methotrexate (FAMTX) prior to surgery or to undergo surgery only. Patients younger than 75 years of age with a good physical and mental condition and a histologically proven adenocarcinoma of the stomach without clinical or radiographic (computed tomography scan) evidence of distant metastases were eligible for this trial. Early gastric cancer or cardia carcinoma were excluded. The response to chemotherapy was evaluated after two and four courses. In case of progressive disease (PD) after two courses, patients were operated upon as soon as possible. Otherwise complete response (CR) partial response (PR) or stable disease (SD), two more courses were scheduled. The standard surgical procedure was a limited lymphadenectomy (D1) with staging biopsy of the para-aortic lymph nodes. Between September 1993 and February 1996, 56 eligible and evaluable patients were entered: 27 were randomised to receive FAMTX before surgery and 29 to undergo surgery only. In the FAMTX + surgery treatment group, 15/27 (56%) had curative resections versus 18/29 (62%) in the surgery only arm. There was no difference in the frequency of TNM stages I + II in both treatment arms: 15/27 versus 15/29. Due to PD and/or toxicity, 12 patients (44%) could not complete the planned four courses of FAMTX. Response evaluation after chemotherapy was possible in 25 patients: 2 CR, 6 PR, 8 SD and 9 PD. The difference in curative resectability rate was 6.5% (95% confidence interval -32 to +19%) in favour of surgery only. Downstaging for stages I + II did not occur. PD was more often the reason for not completing the planned four courses than toxicity. More active regimens than FAMTX are required for future randomised trials.