To improve the refining ability and convergence rate of cuckoo search algorithm for finding optimal solution. An improved cuckoo search algorithm with adaptive method is proposed. The self-adaptive machine is used to control the scaling factor and find probability so as to improve population diversity and avoid premature, as a result, more individuals participating in the evolution, and then refining ability and convergence rate are improved. The result of experiment show the ICS algorithm has better performance when lots of test functions are considered, ICS algorithm has faster convergence speed and higher precision.