Abstract: Cuckoo search algorithm is a new population-based optimization technique inspired by the obligate brood parasitic behavior of some cuckoo species. It searches new solutions by iteratively using Lévy flights random walk and Biased random walk, which employs a mutation and crossover operators respectively. In Biased random walk, the crossover operator with random search schema will be a certain blindness or inefficiency, resulting in weakening the search ability of cuckoo search algorithm. Thus, this paper proposes an orthogonal crossover cuckoo search algorithm with external archive (OXCS). By being embedded in Biased random walk, the orthogonal crossover operator, which is an efficient search schema, is employed to enhance the crossover operator schema so as to polish the search ability of cuckoo search algorithm. The proposed algorithm also utilizes an external archive, which maintains the historical information of population within a certain period, to provide one parent-individual for the orthogonal crossover operator in order to improve the diversity. The comprehensive experiments are carried out on 24 benchmark functions in comparison with other algorithms. The results demonstrate the proposed strategies can improve the search ability of cuckoo search algorithm, and enhance the convergence speed and the solution quality of the algorithm for the continuous function optimization problems effectively.