Objective To evaluate 3.0 T MR DWI techniques in detecting the lesions of pre and post-radiofrequency ablation of the rabbit liver VX2 tumors. Methods Twenty two New Zealand white rabbits were used in this experiment. Twenty tumor fragments were implanted into the livers of 20 rabbits respectively. Two normal rabbits were used as controls for radiofrequency ablation of the normal liver. 3.0 T MR DWI was performed 14 to 21 days after tumor implantation (mean, 17 days) in the tumor-bearing animals. Radiofrequency ablation was performed in the 18 tumor-bearing animals and in the two healthy animals. 3.0 T MRI and DWI were performed 7 to 10 days after radiofrequency ablation (mean, 8 days).Pathology was obtained immediately after the completion of post-radiofrequency ablation MR imaging. The MRI features and ADC values of pre- and post -radiofrequency ablation lesions in the liyers with VX2 tumors and normal rabbits were analyzed and correlation was made with histopathologic findings. Analysis of variance repeated measures were performed in analyzing the differences among the ADC values of different tissues with the same b value. Results All 20 rabbit liver models of VX2 tumors were constructed successfully. One rabbit died of anesthetic overdose, another one showed necrosis within the implanted tumor. All 18 untreated VX2 tumors had predominantly low or iso-signal intensity on T1 WI and high signal intensity on T2WI. All 18 VX2 tumors and 2 normal rabbits were treated by radiofrequency ablation successfully. Lesions treated by Radiofrequency ablation displayed low signal intensity on T1 WI, and high signal intensity on T2WI. Seven to 10 days after radiofrequency ablation, lesions varied from having low signal intensity to slightly increased signal intensity on T1 WI, with areas of mixed ( high, intermediate, and low) signal intensity. A peripheral rim of high signal intensity with varying thickness on T2WI correlated with granulation tissue, which exhibited intense enhancement on contrast-enhanced images. Areas of low to intermediate signal intensity within the lesion on T2WI corresponded to coagulation necrosis. The tumor tissue appeared as areas of peripheral nedularity, with intermediate to high signal intensity on T2-weighted images and DWI. The tumor specimen was gray, among the tumor tissue, there were hyperplastic vessels,and granulation tissue. When b value was 600 s/mm2 , the ADC value of viable tumor (9 cases), necrosis (18 cases), granulation tissue ( 18 cases), normal liver tissue ( 18 cases) were ( 1. 227 ±0. 140) × 10-3,(0. 702 ± 0. 050)×10-3, ( 1.918 ± 0.124) × 10-3, ( 1. 739 ± 0. 044 ) × 10-3 mm2/s, respectively, which were statistically significant (P <0. 01 ). When b =200,400,600,800,1000 s/mm2, the differences of ADC values among viable tumor, granulation tissue, necrosis,normal liver tissue were also statistically significant ( P <0. 01 ). Conclusion The rabbit liver VX2 tumor models and 3.0 T MR DWI are important tools in the basic and clinical researches of radiofrequency ablation. Key words: Liver neoplasms; Magnetic resonance imaging; Radiology,interventional