In this paper, we treat the capacitated location-routing problem (CLRP). The CLRP which combines the facility location problem with the routing problem is one of the most important problems in combinatorial optimisation field. In this study, we propose a new two-stage heuristic (2-SH) algorithm for the large-scale CLRP. In fact, our aim is to find high-quality solutions for very large-scale problems within a short computing time. The proposed method consists of two major stages. In the first, the routing sub-problem is solved using a tabu search. Then, in the second stage, to associate each route to a facility, the location-allocation sub-problem is treated by means of a simulated annealing method. Our proposed approach is evaluated on large-scale randomly generated instances and compared successfully with other two state-of-the-art algorithms from the literature.