The aim of the work is to create and refine methods of algorithmic filter synthesis allowing for the designing of variable structure filters in cases when statistical characteristics of effective signal and interference – non-stationary stochastic processes – are unknown and the application of classic synthesis methods, e.g. Kalman-Busy method, is impossible. The paper presents the basics of an algorithmic variable structure filter synthesis, the essence of which is the formulation of the filter synthesis task in the form of searching optimization and its solution with application of simplex search methods. An example of algorithmic variable structure filter synthesis is presented. Parameters of a variable structure filter transfer function found by algorithmic method correspond to theoretical parameters calculated on the basis of statistical characteristics of input signals.