Despite the existence of a number of variations of genetic algorithms for the traveling salesman problem in literature, no efforts have been made to the best of our knowledge to characterize their performance. This paper presents a detailed comparative study on some of the solution methods for the traveling salesman problem using genetic algorithms. All the operators of genetic algorithms have been given equal emphasis in the analysis. The complete simulation has been done using a number of C programs written by us for this purpose. A permanent network has been particularly considered for this task. The results shed insight on the best solution method for the problem. However, we cannot conclude how the behavior of those algorithms depends on network topology.