To construct a stable rat portal vein thrombosis (PVT) model and explore the time window of urokinase thrombolytic therapy on this basis. Constructing a rat PVT model by combining anhydrous ethanol disruption of portal endothelium with stasis of blood flow. Forty-eight rats after PVT modeling were divided into control group and experimental group, with 24 rats in each group. The experimental and control groups were given urokinase treatment and saline tail vein injection, respectively. The two groups of rats were observed and compared for PVT formation at 1, 3 and 5 days after modeling, respectively. A stable rat PVT model was successfully constructed. No significant differences were found in PVT length, portal vein wet weight, and percentage of luminal occlusion area in the control rats at 1, 3, and 5 days after successful modeling (P > 0.05). Compared with control rats 1 day after modeling, the percentage of non-organized thrombus luminal area was significantly decreased (P < 0.0001), and the percentage of organized thrombus luminal area was significantly increased (P < 0.0001) in the PVTs of control rats at 3 and 5 days after modeling. After thrombolytic treatment with urokinase, plasma fibrinogen (FBG) levels were significantly decreased in the experimental group of rats compared with the control group (P < 0.0001), and plasma D-dimer (D2D) levels were significantly increased in the experimental group of rats compared with the control group (P < 0.0001). In addition, we observed prolongation of prothrombin time (PT) in the experimental group at 1, 3 and 5 days after modeling compared to the control group (P = 0.0001). Compared with the control group, portal vein wet weight and PVT length were significantly decreased in the experimental group of rats at 1 day after modeling (P < 0.05), whereas these differences were not found in the two groups of rats at 3 and 5 days after modeling (P > 0.05). The percentage of non-organized thrombus area in the experimental group was significantly decreased compared with that in the control group at 1, 3, and 5 days after modeling (P < 0.05), whereas there was no significant difference in the percentage of lumen area of organized thrombus between the two groups (P > 0.05). The method of producing a rat PVT model by destroying the endothelium of the portal vein by anhydrous ethanol combined with blood flow stasis is feasible and reproducible. In addition, the optimal time window for thrombolysis in the treatment of PVT in rats using urokinase is the early stage of thrombosis, when the fibrin content is highest.