To determine, in vivo, the effect of radiofrequency ablation (RFA) treatment time and tissue blood flow on the size and shape of the resulting necrotic lesion in porcine liver. Radiofrequency ablation is an electrosurgical technique that uses a high frequency alternating current to heat tissues to the point of desiccation (thermal coagulation). Radiofrequency ablation is well established as the treatment of choice for many symptomatic cardiac arrhythmias because of its ability to create localized necrotic lesions in the cardiac conducting system. Until recently, a major limitation of RFA was the small lesion size created by this technique. Development of bipolar and multiple-electrode RFA probes has enabled the creation of larger lesions and therefore has expanded the potential clinical applications of RFA, which includes the treatment of liver tumors. A basic understanding of factors that influence RFA lesion size in vivo is critical to the success of this treatment modality. The optimal RFA technique, which maximizes liver lesion size, has yet to be determined. Theoretically, lesion size varies directly with time of application of the RF current, and inversely with blood flow, but these relationships have not been previously studied in the liver. Six animals underwent hepatic RFA (460 kHz), for 5, 7.5, 10, 12.5, 15, and 20 minutes. Identical, predetermined anatomic areas of the liver were ablated in each animal. Two additional animals underwent 12 RFA treatments -- 6 with vascular inflow occlusion (Pringle maneuver) and 6 with uninterrupted hepatic blood flow. Animals were euthanized and the livers were removed for gross pathologic examination. All lesions were measured in three dimensions and photographed. Tissues were examined by routine histology and by histochemistry to determine viability. Increasing duration of RFA application from 5 through 20 minutes did not create lesions of larger diameter, but this time increase did predict deeper lesion production (beta = 0.34, p = 0.04). A range of lesion shapes were created from four separate ovals (corresponding to each electrode), to larger ovals intersecting to form a cross, to spheroid lesions. The number of blood vessels in close proximity to the probe tip (within a 1-cm radius from the center of the lesion) strongly predicted minimum lesion diameter (beta = -0.61, p = 0.0001) and lesion volume (beta = -0.56, p = 0.0004). This negative effect of blood flow on lesion size was confirmed experimentally. Radiofrequency ablation lesions created during a Pringle maneuver were significantly larger in all three dimensions than lesions created without a Pringle maneuver: minimum diameter was 3.0 cm (with Pringle) versus 1.2 cm (p = 0.002), maximum diameter was 4.5 cm (with Pringle) versus 3.1 cm (p = 0.002), depth was 4.8 cm (with Pringle) versus 3.1 cm (p < 0.001), and lesion volume was 35.0 cm3 (with Pringle) versus 6.5 cm3 (p < 0.001). Blood flow is a strong predictor of all RFA lesion dimensions in porcine liver in vivo, whereas a change of treatment time from 5 to 20 minutes is predictive only of lesion depth, but not diameter or volume.