A novel evolutionary optimization method, clustering searching algorithm (CSA), is presented. In CSA, a virtual cluster group is constructed among individuals in order to adjust the operation state of simulated evolutionary system dynamically and improve the searching efficiency of population. After introducing the basic model of CSA, this paper presents CSA/DE, a new CSA blending the evolutionary search operators with the differential computing mechanism for solving numerical optimization problem. In simulations, 6 classical multidimensional functions and 6 challenging composition functions are selected to test the performance of CSA/DE. The experimental results show CSA/DE is an efficient and reliable search optimization algorithm for multidimensional continuous problem. The work of this paper verifies the feasibility and validity of CSA. Meanwhile, this research demonstrates, based on CSA, multiple heterogeneous searching mechanisms can be merged into one algorithm to get much more pertinence and harmony in searching process, thus providing a viable way for designing high-performance optimization algorithm.