Abstract: The strong and weak points of direct-difference, variational calculus and direct-adjusting divergence methods for mesoscale wind field have been evaluated by numerical experiments. The major results of experiments are as follows: (1) As shown in other experiments, the direct-difference method can get a wind field of minimum integral divergence with a short computer time, but the vertical motion obtained in the upper calculation area is overestimated. (2) The variational calculus method is able to overcome the above disadvantage, but the computer time needed is too much. (3) The direct-adjusting divergence method possesses the advantages of the former two methods and can get a reasonable divergence wind field with a compromising computer time.