This paper considers the problem of the guaranteed generalized H-2 performance state estimation for delayed static neural networks. The activation function is assumed to satisfy a sector-bounded function. Based on a suitable Lyapunov-Krasovskii functional and a new integral inequality, an improved delay-dependent criterion is presented such that the error system is globally asymptotically stable with guaranteed generalized H-2 performance. Compared with some existing results, the presented sufficient condition can provide much better performance. A numerical exampl...