Three kinds of constrained traveling salesman problems (TSP) arising from application problems, namely the open route TSP, the end-fixed TSP, and the path-constrained TSP, are proposed. The corresponding approaches based on modified genetic algorithms (GA) for solving these constrained TSPs are presented. Numerical experiments demonstrate that the algorithm for the open route TSP shows its advantages when the open route is required, the algorithm for the end-fixed TSP can deal with route optimization with constraint of fixed ends effectively, and the algorithm for the path-constraint could benefit the traffic problems where some cities cannot be visited from each other.