In this paper, a new computing environment, so called ``Parallel Computing`` is adopted to get the numerical results of CFD problems. To apply the traditional CFD algorithm into the parallel computer, the domain decomposition technique compiler dependent technique and PVM message passing library are used. By distributing computing loads, we can get reasonable computing time to solve CFD problems with the unstructured grids. As a result, the possibility of massive usage of parallel computers in near future is demonstrated.