We present a technique to model the instrumental polarization in the focal plane of a telescope. It takes into account that different rays of an incoming beam suffer different variation of the original polarization in their paths through the system. It also considers that the net effect depends on the way in which the different rays interfere with each other. We show that the Mueller matrix which describes this instrumental polarization greatly simplifies if the polarimetric measurements have poor spatial resolution. The Mueller matrices corresponding to a pair of academic cases are worked out: a diffraction limited telescope when the source lies on its axis and an axisymmetric mirror with the source off the axis. We discuss the consequences of these mathematical results on real telescopes (e.g. LEST). Finally, we briefly consider the seeing-induced instrumental polarization.