This paper describes the state of the art of agent-based approach to simulate social and economic complex systems. The approach is characterized by software agents with individual internal models, their interactions in a given environment, and the micro-and macro links among the individual behaviors and social phenomena. Using agent-based simulation, thus, we are able to analyze complex phenomena in the real world, which is difficult to understand via the other approaches. This paper examines the literature, compares the model with traditional ones, explains some of our recent results, and explores the future perspectives.