To establish models of portal vein occlusion of hepatic VX2 tumor in rabbits and to evaluate the value of multi-slice CT. Forty New Zealand rabbits were divided into 4 groups according to digital table: Immediate group (group A; transplantation of tumor immediately after the portal vein occlusion), 3-wk group (group B; transplantation of tumor at 3 wk after the portal vein occlusion), negative control group (group C) and positive control group (group D), 10 rabbits in each group. Hepatic VX2 tumor was transplanted with abdominal-embedding inoculation immediately after the portal vein occlusion and at 3 wk after the portal vein occlusion. Meanwhile, they were divided into negative control group (Left external branch of portal vein was occluded by sham-operation, and left exite was embedded and inoculated pseudoly) and positive control group (Transplanted tumor did not suffer from the portal vein occlusion). All rabbits were scanned with multi-slice CT. All 40 animals were employed in the final analysis without death. Tumor did not grow in both immediate group and 3-wk group. In 3-wk group, left endite was atrophied and growth of tumor was inhibited. The maximal diameter of tumor was significantly smaller than that in positive control group (2.55 +/- 0.46 vs 3.59 +/- 0.37 cm, t = 5.57, P < 0.001). Incidences of metastasis in the liver and lung were lower in 3-wk group than those in positive control group (10% vs 40%, and 90% vs 100%, respectively). The expression intensities of the vascular endothelium growth factor (VEGF) in groups A, B, C and D were 0.10 +/- 0.06, 0.66 +/- 0.21, 0.28 +/- 0.09 and 1.48 +/- 0.32, respectively. VEGF expression level in the test group A was significantly lower than that in the negative control group C (t = 5.07; P < 0.001). In addition, VEGF expression in the test group B was significantly lower than that in the positive control group D (t = 6.38; P < 0.001). Scanning with multi-slice CT showed that displaying rate of hepatic artery branches was obviously lower in grade III (40%) than that in grade I (70%) and II (100%) (P < 0.05); but there was no significant difference in displaying rate of the portal vein at various grades. Values of blood flow (BF) of the liver, blood volume (BV), mean transit time (MTT) and permeability of vascular surface (PS) were lower in the immediate group and 3-wk group than those in control groups, but values of hepatic arterial fraction (HAF) were increased. Significant positive correlations were existed between BF and BV (r = 0.905, P < 0.01), and between BF and PS (r = 0.967, P < 0.01), between BV and PS (r = 0.889, P < 0.01). A significant negative correlation existed between PV and HAF (r = -0.768, P < 0.01), between PS and HAF (r = -0.557, P < 0.01). The values of BF, BV and PS had a positive correlation with VEGF (r(BF) = 0.842, r(BV) = 0.579, r(PS) = 0.811, P < 0.01) . However, there was no significant correlation between the values of MTT and HAF and the VEGF expression (r(MTT) = 0.066, r(HAF) = -0.027). Ligating the left external branch of portal vein is an ideal way to establish models of portal vein occlusion in rabbits with hepatic VX2 tumor. Multi-slice CT plays a key role in evaluating effect of portal vein occlusion.