Reduced osmolarity in incubation medium was previously shown to increase in vitro cellular accumulation and cytotoxicity of cisplatin on cancer cells. We confirmed in the present work that cisplatin diluted in an hypotonic 25 g/l glucose solution (124 mOsm) was dramatically more cytotoxic in vitro than cisplatin diluted in normotonic 9 g/l NaCl (300 mOsm) on the human HT29 colon and MCF7 breast cancer cells. We conducted then a pilot clinical study on the administration of cisplatin diluted in hypotonic 25 g/l glucose solution given through the balloon-occluded hepatic artery for the treatment of liver metastases from colon or breast cancer tumors. Nine patients (5 men, 4 women; mean age 58, range: 44-71) with confirmed isolated, unresectable metastases from colorectal (7) or breast (2) tumors were included in this study and a total of 23 cycles were administered (2.55 per patient; range 1-5) with an average dose of 50 mg cisplatin (range: 12.5-100). Hepatic artery dissection due to balloon injury with partial or complete arterial obstruction were encountered in 2 patients. Pain in the liver and epigastric area was the main symptom which was constant and intense during the IAH cisplatin injection. Fever > 38 degrees C was observed in 15/23 cycles and increase of creatinine in 1/23 cycles. Transient increase of hepatic transaminases without change in prothrombin time was registered in all patients. However one patient who received the highest dose of 100 mg cisplatin developed a persistent but reversible clinical jaundice and a transient increase in prothrombin time. One patient achieved a partial response (12 weeks), 7 had stable disease (mean duration: 6 weeks) and one had a progressive disease. Hepatic arterial infusion of cisplatin diluted in hypotonic 25 g/l glucose solution and administered through the balloon-occluded hepatic artery is a feasible approach. Total dose of cisplatin in hypotonic glucose solution will not exceed 80 mg by cure in a further phase II study.