Abstract: Considering the multi-level distribution network has becoming more and more common, a two-echelon location routing problem (2E-LRP) model was established based on minimum total cost objective function. To solve the 2E-LRP model, a simulated annealing with large neighborhood search algorithm was developed. In the framework of the simulated annealing algorithm, a large neighborhood search process was embedded, which includes destroy-and-repair principles as well as some local search methods to further improve the range of the neighborhood search in the solution space. The proposed model and algorithm were tested by two-echelon benchmark instances and compared with the standard simulated annealing algorithm solutions and the internationally best known solutions. The results show the proposed model and algorithm to be correct and that the algorithm can obtain better solutions than standard simulated annealing when solving large-scale problems.