Abstract: This paper is concerned with the problem of passivity analysis for stochastic neural networks with mixed delays.It is assumed that the discrete delay is uncertain and seen as a composition of a nominal value subject to a time varying perturbation.The distributed delay is supposed to be constant.By constructing a appropriate discretized Lyapunov-Krasovskii functional and utilizing free-weight matrix techniques,delay-dependent passivity criteria are established in terms of linear matrix inequalities to guarantee the passivity per-formance of the addressed neural networks.