In a randomized clinical trial, gemcitabine (GEM) was more effective than 5-fluorouracil (5-FU) in advanced pancreatic cancer patients. GEM and 5-FU have different mechanisms of action and their combination, from a theoretical point of view, could result in a higher activity. To test activity and feasibility of such a combination, a multi-institutional phase II study was initiated in November 1996 by the Italian Group for the study of Digestive Tract Cancer (GISCAD). Primary objectives of this study were to determine the activity in terms of response rate and clinical benefit, while the secondary objective was toxicity. According to the optimal two-stage phase II design, 54 patients were enrolled. Schedule was: GEM 1000 mg m(-2) intravenous (i.v.), and 5-FU 600 mg m(-2) bolus i.v. weekly for 3 weeks out of every 4. All the 54 patients were symptomatic (pain, weight loss, dyspepsia). A clinical benefit was obtained in 28 patients (51%) (95% confidence interval (CI) 38-64%). Two patients achieved a partial response and 34 a stable disease. Median survival for all the patients was 7 months. Side-effects were mild: no gastrointestinal or haematological grade 3-4 toxicity (WHO) were recorded. We observed only six episodes of grade 2 (WHO) leukopenia and seven episodes of thrombocytopenia. Although the non-randomized design of this study suggests caution in the interpretation of these data, in consideration of the low incidence of toxicity and the favourable results obtained in terms of clinical benefit, it may be worthwhile to test more active schedules of 5-FU (continuous infusion) in combination with gemcitabine.