In this paper, a new method for state estimation, referred to as the variable structure filter (VSF), is briefly reviewed. The VSF method is model based and has been formulated for its application to linear systems. It provides a means of explicitly defining the level of uncertainty in the dynamic model used by the filter, thus allowing for tradeoff between the performance indicators of the filter. This trade-off feature is not generally explicitly available in some of the more established concepts used for state estimation. In this paper, an extension to the VSF method for its application to nonlinear systems is proposed and referred to as the extended variable structure filter (EVSF). The derivation of EVSF and its application to a nonlinear robotic example is provided.