Gastrointestinal toxicity from hepatic arterial infusion (HAI) of floxuridine in patients with liver metastases is probably due to extrahepatic perfusion or to partial escape of the drug from first-pass liver extraction. The aim of this study was to verify the role of technetium-99m-labelled macroaggregated albumin (99mTc-MAA) arterial catheter perfusion scintigraphy at the beginning of each chemotherapy cycle in decreasing or preventing gastrointestinal toxicity. We studied 167 consecutive patients. On the basis of the scintigraphic follow-up and the presence or absence of an intrahepatic arteriovenous shunt (IHAVS), we classified our patients into the following groups: (1) FU+ hepatic distribution pattern (DP), comprising 29 patients with regular scintigraphic follow-up who showed the expected distribution pattern at each control or a distribution pattern with transient alterations (extrahepatic escape) promptly reversed by the replacement of the catheter. Among these 29 patients there was one case of gastrointestinal toxicity. (2) FU- hepatic DP, comprising 128 patients who were evaluated with 99mTc-MAA only at the beginning of the first chemotherapy cycle, showed the expected distribution pattern and underwent HAI with no further scintigraphic evaluation. Among these 128 patients there were 28 cases of gastrointestinal toxicity. (3) FU+ pulmonary DP, comprising three patients with abnormally elevated pulmonary uptake (higher than 5%) and with regular scintigraphic follow-up. There were two cases of gastrointestinal toxicity among these three patients. (4) FU- pulmonary DP, comprising seven patients with abnormally elevated pulmonary uptake and without regular scintigraphic follow-up. There were four cases of gastrointestinal toxicity among these seven patients. The incidence of toxicity was significantly higher in group FU- hepatic DP than in group FU+ hepatic DP (21.9% vs 3.4%, P<0.05). In both the FU+ pulmonary DP and FU- pulmonary DP groups, the incidence of gastrointestinal toxicity was higher than 50%, with no significant difference between them. We conclude that, when performing 99mTc-MAA perfusion scintigraphy, the presence of an abnormally elevated pulmonary uptake (IHAVS higher than 5%) is the most relevant positive prognostic index for the development of gastrointestinal toxicity. Furthermore, in the absence of abnormal pulmonary uptake (IHAVS lower than 5%), strict scintigraphic follow-up is useful since it is able to promptly diagnose the presence of extrahepatic abdominal perfusion and thus to prevent the occurrence of gastrointestinal toxicity.