Abstract: An LPT (Largest Processing Time) approximate algorithm for the load balancing problem in CFD parallel computation with structural mesh is developed based on a scheduling method. Using the theory of greedy strategy, the feasibility and limitations of applying LPT approximate algorithm to the load balancing problem are analyzed, and an LPT improved algorithm is proposed. Then, numerical simulations are implemented to evaluate the performances of the LPT approximate algorithm and the LPT improved algorithm. Computational results show that the LPT approximate algorithm can be applied to solve the load balancing problem in CFD parallel computation, and the LPT improved algorithm developed in this paper provides a more balanced load allocation than the LPT approximate algorithm does.