The presented paper shows a possibility of creation of a useful and efficient model of conductivity for composite polymeric electrolytes. It deals with a more elaborate and versatile mesoscale approach (in comparison to the already applied Effective Medium Theory approach for which each new composite microstructure forces a new mixing rule development) based on the Random Resistor Network method. The introduced model can predict the ac conductivity values for a whole spectrum of frequencies for different composites with a wide range of possible microstructures. It is based on complex current iteration procedure for a randomly generated three-dimensional network of impedances constituting a digital image of the electric properties of the virtual sample studied. A spatial unit of the material present in the sample is defined as a parallel RC connection. (c) 2007 Published by Elsevier B.V.