Data distribution is one of the key factors that affect the performance of programs running on distributed memory multiprocessors. This paper presents a 0-1 integer programming based approach for effective data distribution using data transformation techniques. This approach uses data transformations to change memory layouts and hence makes effective data distribution possible. Moreover, it also uses data distribution graphs to describe how programs are parallelized and how arrays are accessed, and transforms global data distribution problems into the problems of finding optimal paths in data distribution graphs. Therefore, 0-1 integer programming can be used to solve optimal path problems. The approach can effectively distribute multidimensional arrays with affine subscripts accessed in multiple loop nests and can exploit the parallelisms of loop nests as much as possible. In addition, it can also solve offset alignment problems. Thus data communication overheads can be reduced as much as possible. The experimental results show that the approach presented in this paper is effective.