A novel Chaotic Improved Cat Swarm Algorithm (CCSA) is presented for global optimization. The CSA is a new meta-heuristic optimization developed based on imitating the natural behavior of cats and composed of two sub-models: tracing mode and seeking mode, which model upon the behaviors of cats. Here different chaotic maps are utilized to improve the seeking mode step of the algorithm. Seven different chaotic maps are investigated and the Logistic and Sinusoidal maps are found as the best choices. Comparing the new algorithm with the CSA method demonstrates the superiority of the CCSA for the benchmark functions.