This paper deals with the exponential H∞ filtering problem for discrete-time neural networks switched singular systems with time-varying delays. The main purpose of this paper is to design a linear mode-dependent filter such that the resulting filtering error system is regular, causal, and exponentially stable with a prescribed H-infinity performance bound. In addition, the decay rate of the filtering error dynamics can also be tuned. By constructing an appropriate Lyapunov functional together with some zero inequalities and using the average dwell time scheme, a novel delay-dependent sufficient condition for the solvability of the H-infinity filtering problem is derived. Based on this condition, the desired filter gains can be obtained by solving a set of linear matrix inequalities (LMIs). A numerical example is presented to show the effectiveness of the proposed design method.