Patient dosimetry data for intra-arterial()iodine-131 lipiodol therapy for hepatocellular carcinoma (HCC) are scarce. The aim of this study was to determine the absorbed dose (D) to the tumour and healthy tissues, as well as the effective dose (E), by different methods for 17 therapies in 15 patients who received a mean activity of 1.9 GBq (SD 0.2) (131)I-lipiodol. Eight patients received thyroid blocking by potassium iodide (KI). Patient dosimetry was performed based on bi-planar total body scans using the Monte Carlo simulation program MCNP-4B and the MIRDOSE-3 standard software program. CT images of each patient were used to determine liver and tumour volume and position. The total body dose to the patient was also determined by biological dosimetry with the in vitro micronucleus (MN) assay. From the increase in micronucleus yield after therapy, the equivalent total body dose (ETBD) was calculated. Results for D and E were comparable between MCNP and MIRDOSE (liver: mean 7.8 Gy, SD 1.8, lungs: 6.8 Gy, SD 2.9, E: 2.01 Gy, SD 0.58). MIRDOSE gave a systematic overestimation for the tumour dose, especially for tumours <3 cm (15%). The MCNP method is more accurate since the dose contributions from tumour to organs and vice versa can be accounted for. The absorbed dose to the thyroid was significantly lower for patients who received KI (7.2 Gy, SD 2.2) than for the other patients (13.8 Gy, SD 5.0). MN yields could be obtained for only 12 of the 17 therapies due to hypersplenism. A mean ETBD of 1.66 Gy (SD 0.73) was obtained, but the MN results showed no correlation between the ETBD and the total body dose values of the physical dosimetry. Also, in all except one of the patients, no further reduction in the number of thrombocytes was observed after therapy, probably due to the existing hypersplenism. It is concluded that in view of the high E values, patient dosimetry is necessary for patients receiving (131)I-lipiodol therapy. Except in the case of the smaller tumours, comparable results were obtained with MCNP and MIRDOSE. Due to hypersplenism, biological dosimetry results based on the MN assay are not reliable.