Abstract: In this study, we compare three meta-heuristics approaches: hill-climbing, simulated annealing and late acceptance hill-climbing algorithm, for solving traveling salesman problem. All algorithms were tested on seven traveling salesman problem instances listed in LIBTSP datasets. Each algorithm was implemented and evaluated independently using its parameter. Performance were compared between all algorithms and evaluated in terms of objective function that calculated to find the optimal rout within several selected routes. The main conclusion is that late acceptance hill-climbing algorithm performs better than simulated annealing and simple hill-climbing as far as solution quality is concerned whilst simple hill-climbing gives the worst performance according to its poorness in global search.