Simulated annealing (SA) has been applied with success to many numerical and combinatorial optimization problems in recent years. SA has a rather slow convergence rate, however, on some function optimization problems. In this paper, by introducing chaotic systems to simulated annealing, we propose a optimization algorithm named chaos simulated annealing (CSA). The distinctions between CSA and SA are chaotic initialization and chaotic sequences replacing the Gaussian distribution. Simulation results of typical complex function optimization show that CSA improves the convergence and is efficient, applicable and easy to implement. In addition, we discuss the advantages of CSA, and show the reasons why CSA performs better than SA.