Considering that the existing central force optimization (CFO) cannot achieve an effective balance between the evolution speed and the quality of solutions, an improved central force optimization based on the simplex method (SM-CFO) was introduced. By periodical migration of the best individual obtained by the SM operator into the detector population of the CFO, the proposed algorithm can achieve cooperative search of the CFO and SM: with the help of CFO, SM can get away from local minima; and with SM, CFO can improve its local exploiting capability. Furthermore, in order to enhance the ability of CFO and SM, an improved Nelder-Mead SM was proposed. Through a detailed sensitivity analysis on the parameters of the proposed algorithm, some suggestions for the parameter setting were put forward. Numerical experiments and comparisons on six 2-40 dimensional benchmark functions indicate that the proposed algorithm avoids the stagnation and enhances the global search ability, and is superior to other existing algorithms.