AIMS: Chemotherapy as a palliative therapy option for patients with advanced pancreatic cancer remains disappointing. Some authors have recently claimed high response rates and a prolongation of median survival after regional chemotherapy. Isolated perfusion may result in the highest drug concentrations within the target tissue without causing systemic side-effects. An established, commercially available system of isolated hypoxic perfusion (IHP) was therefore evaluated in patients with unresectable or recurrent pancreatic cancer. PATIENTS AND METHODS: IHP was performed in 17 patients with unresectable pancreatic cancer. Five women and 12 men with a median age of 61 years were treated. A 20-min isolated hypoxic perfusion was performed after 40 mg of mitomycin C (MMC) had been instilled into the running perfusion system over 5 min. Tumour response was evaluated by CT-scan 6 weeks after IHP. RESULTS: Twenty perfusions were carried out in 17 patients. Within 10 min of perfusion, the perfusate's PO2 decreased to 13% of the baseline value. Five minutes after the infusion of MMC a local concentration of 15.2 mg/litre was observed. Toxicity-related deaths did not occur. Nausea and vomiting (NCI> or =II: 12 episodes) were the most frequently observed toxicities after IHP. In five patients (29%) a deep vein thrombosis occurred. None of the treated patients responded to the regimen used in this trial. The median survival time after IHP was 4.2 months (range 1.3-21). CONCLUSIONS: The overall rate of side-effects and complications after IHP was high. In spite of some hopeful reports on this treatment in recent years, IHP did not show any benefit in terms of tumour response or median survival. On the basis of these experiences, this procedure should no longer be used as treatment for patients with unresectable or recurrent pancreatic cancer.