Detecting communities is of great importance in complex networks, which is very hard and not yet satisfactorily solved. This paper investigates a backbone guided search algorithm in order to optimize the modularity of the community structure in complex networks. Our proposed algorithm consists of two main procedures, which are local search and backbone guided search. When the local search procedure can not improve the modularity function any more, the backbone guided search procedure is applied for further improvements. The computational results indicate the effectiveness of the proposed algorithm compared with the classical methods.