Treatment parameters of extracorporeal high intensity focused ultrasound (HIFU) were analysed in normal and tumor-bearing rabbit liver. HIFU was generated with a 1 MHz transducer and energy was provided by a 7.5 kW power amplifier. In vivo experiments were conducted on 74 New Zealand rabbits. Normal rabbits and rabbits bearing an intrahepatic VX2 tumor were used. In group 1, spatial peak temporal peak (SPTP) intensities ranging from 1470 to 5500 W cm-2 and exposure times from 0.5 to 5 s were tested at a constant depth in the liver; in group 2, the output power was adjusted as a function of the target depth in order to keep constant the focal in situ intensity in the liver; in group 3 (liver tumors), the focal in situ intensity was 1365 W cm-2 in eight rabbits and 500 W cm-2 in nine. In groups 1, 2 and 3, rabbits were sacrificed 48 h after the treatment. Groups 4 and 5 were designated for analysis of the lesion in the normal liver 4 weeks after treatment at 1000 W cm-2 and 3000 W cm-2 SPTP intensities, respectively. In normal rabbits, the lesion volume increased with exposure time at constant intensity; there was a negative correlation between intensity and exposure time (group 1). When the output power was adjusted as a function of the path length, the lesion size was nearly constant (group 2). In VX2 rabbits, tumor destruction rates were significantly higher in rabbits treated at 500 W cm-2 than in rabbits treated at 1365 W cm-2 (p < 0.05; group 3). As in the normal liver, the lesion volume increased with the exposure time at constant intensity. HIFU lesions treated at 1000 w cm-2 (SPTP) healed as thin fibrous scars, and no severe complication occurred (group 4); at 3000 W cm-2 (SPTP), scars were larger and perforation of a neighbouring organ was seen in 7 of 11 rabbits (group 5).