Abstract: This paper is concerned with a class of extended bidirectional associa-tive memory (BAM) neural networks with delays and impulses. By means of fixed point principle in Banach space and Lyapunov function, sufficient conditions are obtained for the existence and global exponential stability of a unique equilibrium without assuming the activation function to be bounded, monotonic or differentiable. An illustrative ex-ample is given to demonstrate the effectiveness of the obtained results.