Four patients with advanced hepatic malignancy have been investigated with a view to intra-arterial treatment with degradable starch microspheres (DSM) and mitomycin-C (MMC). Computer-assisted interpretation of data gathered during the administration of technetium-labelled albumin allows the degree of arteriovenous shunting within the liver to be measured. In a similar way, the effectiveness of DSM in blocking hepatic arterial flow is determined. Two patients showed unacceptably high levels of shunting (75% and 100%) and were excluded from further study. One patient with hepatocellular carcinoma shunted only 24% of the macroaggregated albumin, and flow through the liver was reduced to 48% by DSM therapy. This patient has undergone four courses of MMC without significant toxic complications. The last patient had 25% shunting. The DSM titration and the first MMC administration had to be abandoned because of severe abdominal pain and nausea related to a fracture of the intra-arterial catheter. Complications directly associated with the procedures have been acceptable. The relationship between flow reduction and the amount of DSM administered is non-linear, and the appropriate dose of DSM for each treatment must be titrated against both the patient's symptoms and the computer-generated flow indices.