Abstract: The cuckoo search algorithm (CSA) is a new nature-inspired algorithm which is simple and efficient. This algorithm uses Lévy flights random walk and biased random walk to search for new solutions. Aiming at increasing the performance of cuckoo search algorithm via improving parameter, this paper presents Adaptive Cuckoo Search Algorithm with Memory Strategy(MACS). By adjusting Lévy flight adaptively based on premature convergence degree and the fitness of the cuckoo, the algorithm makes a compromise between global search and local search.The ability of local search has improved via introducing memory strategy.Simulation results have showed that the performance of MACS is obviously improved in the aspect of the convergence and accuracy compared with the CS algorithm and other improved editions.