Objective To generate 2-dimensional (2D) vascular networks by numerical simulation of tumor angiogenesis moving along random directionin order to provide more realistic microvascular networks for the study of hemodynamics, drug delivery and anti-angiogenesis in solid tumors. Methods To extend the migration of endothelial cells on sprouts from four directions-moving up, down, left or right-in previous models to random direction in the present model, and develop a relevant 2D discrete model to simulate the process of tumor angiogenesis and compare the simulation results with some corresponding experimental data. Results Contrast with the unimproved or some previous models, the features of tumor microvascular networks generated from our model by simulating the angiogenesis in caecum tumor, such as vascular tortuosity, branching and anastomosis, are closer to the physiological facts and more resemble the experimental observations. Conclusions The present model could simulate the process of tumor angiogenesis numerically, and provide a relatively more actual network structure of tumor microvasculature for the research about microcirculation, drug delivery and anti-angiogenesis in solid tumors.