Objective: The very limited efficacy of current chemotherapeutic strategies in advanced pancreatic cancer (APC) and the pattern of metastastic spread, largely confined to the upper abdominal organs within the arterial supply of celiac axis, induced us to design this phase-II study of locoregional intra-arterial chemotherapy. The aim of the present study was to evaluate the feasibility, the toxicity, the response rate and survival of a new combination of drugs administered intra-arterially in the treatment of APC. Methods: From January 1994 to August 1995, thirty-six consecutive patients with APC were given intra-arterial cycles of chemotherapy every 3 weeks through a catheter in the celiac axis introduced via the femoral artery. Seventeen patients were classified as UICC stage III and 19 as stage IV. Nineteen had liver metastases and 1 patient also had lung metastases. The schedule was: 5-fluorouracil 1,000 mg/m2; folinic acid 100mg/m2; epirubicin 60 mg/m2 and carboplatin 300 mg/m2. Each drug was infused over a period of 10 min and only 1 day of hospitalization was necessary for each cycle. After 3 cycles, when a response or stable disease had been obtained, another 2 cycles were planned, health conditions permitting. Results: A total of 114 courses of chemotherapy were administered with a mean of 3 for each patient (range 1-5). 33 patients were evaluable for response by CT scan: 7 of 33 (21%) had a partial response; 16 of 33 (49%) had stable disease; 10 of 33 (30%) had a progressive disease, and 23 of 36 (64%) had a decrease in Ca 19-9. A reduction in pain was reached in 16 of 26 patients (62%) and particularly 9 of 26 patients (35%) showed complete regression for 8 weeks or more. Weight increase was obtained in 13 of 28 patients (46%). Grade-3-4 hematological toxicity was observed in 9 of 36 (25%), and ematemesis in 2 of 36 patients (6%). Grade-3 gastrointestinal toxicity was observed in 4 of 36 (11%); alopecia in 3 of 36 (8%). One sudden death was observed in a patient on day 23 after the third cycle. No complication related to the angiographic procedure was noted. At a median follow-up of 7 months (range 1-17), the median survival was 6.2 months and, according to stage, 4.9 and 13.4 months for stages IV and III, respectively. The median time to progression was 4 months (range 2-11). Conclusions: This study shows that this drug combination, given through a celiac axis infusion, is well tolerated and active, requires only 1 day of hospitalization and might become an interesting form of integrated strategy in a palliative setting. Stage-111 patients are probably the group that might profit most with prolonged survival.