Abstract: This article summarizes the characteristics of chaotic dynamics, reviews the research progresses of chaotic neuron model and chaotic neural networks in recent years. On the basis of them, two kinds of chaotic neural networks are introduced, and their structures and characters are analyzed. The research results show that the chaotic neural networks are more effective than other existing neural networks to solve associative memory and combination optimization problems. Finally, this article presents the fields where chaotic neural networks can be applied and should be further investigated.