Abstract: The study deals with the efficiency of the parallel computation of the Travelling Salesman Problem (TSP) using the genetic algorithms and an unsupervised fuzzy clustering. First, the cities are classified by a clustering algorithms. Second, each class of cities is considered as a sub-tour TSP problem. A parallel genetic algorithms is used for solving the sub tour TSP problem. The main aim by creating the parallel algorithm is to accelerate the execution time of solving TSP. A connection method is proposed to connect the sub-tours into a global tour of whole cities. Furthermore, this global tour is resolved by genetic algorithms for cluster centers and a heuristic scheme. Experimental results, on Master Slave architecture with different TSP problemes show the efficacy of the proposed algorithm in parallelism exploitation.