Cell formation which involves grouping similar parts into part families and the corresponding machines into machine cells is one of the key research issues for designing cellular manufacturing system. Functionally identical machines may vary in production performance. In the paper, processing cost is employed to differentiate the performance of machines, and a two-stage cell formation approach for manufacturing systems with multiple functionally identical machines is proposed. With this approach, parts are firstly clustered into part groups based on process similarity; secondly, machines are then assigned to part groups according to their performance to form manufacturing cells. A heuristic algorithm based on process similarity is presented for part group identification, and a mathematical model with the objective of minimizing the total processing cost is established. An example is employed to illustrate the benefit of the proposed approach and the results suggest that the proposed approach is feasible.