This paper basically reviews the development of regional science and emphasizes the fact that today more and more researchers gradually give up the traditional ideas, begin to treat region as an open, evolutionary, non-linear complex system. The research tendency of regional science has changed from static, equilibrium paradigm to dynamic, evolutionary paradigm, which makes use of sciences of complexity to study and analyze the regional problems and its rules of spatialtemporal evolution. The existence of a vast literature to a certain extent suggests that regional growth as well as its spatial evolution is more complicated than what we had thought. It is indeed the complexity inherent in the regional process that deserves further study. Urban and regional spatial structure can be seen as a cumulative on temporal dimension and agglomeration on spatial dimension which results from numerous interaction and locally decision of adaptive agents (households, enterprises, government and other organizations). Therefore the basic force driving the evolution of spatial structure is inherently microscopic. Our research on spatial complexity should focus on the microscopic mechanism of macroscopic emergence and its simulation. Moreover, this paper also discusses the other primary issues of spatial complexity and potential ways of approaching these problems such as multi-agent model based on the theory of complex adaptive systems.