This paper addresses the issue of robust filtering for time varying uncertain discrete time systems. The proposed robust filters are based on a regularized least-squares formulation and guarantee minimum state error variances. Simulation results indicate their superior performance Over other robust filter designs.