Computational Fluid Dynamics (CFD) is one of the eld which can fully utilize the capacity of existing HPC systems. There are many cases either from basic or applied research which are so complex that their numerical simulation with requested accuracy requires very ne representation of the computational domain. To solve certain problems numerical models consisting of hundred billions of cells are necessary. There are several approaches to create such huge meshes. One of them is based on global mesh re nement and is also known as mesh multiplication. This approach was already described in [1, 2]. Global re nement was already implemented into Code Saturne enhancing its capability in terms of mesh re nement. Meshes with sizes of up to one hundred billion of cells were generated on the y. Since there are many CFD problems where only local area is of interest (either areas close to boundaries, small geometrical entities or in regions with high gradient of solved quantities), local re nement is another approach for mesh creation. In this paper implementation of parallel local re nement applied to Code Saturne is described. The bottleneck of local adaptive re nement is that it breaks load balancing of the original mesh and requires a lot of global communications. Strategy to re-partition the mesh before its re nement is a key issue for optimal resource utilization. To minimize the amount of data transferred among cores it is necessary to do most of the communication during the preprocessing step on the coarse mesh before re nement. Local mesh re nement strategy was tested and its scalability and performance within Code Saturne were analysed. Results are presented in this paper.