In cellular networks, the base stations are usually assumed to form a lattice or (more recently) a Poisson point process. In reality, however, they deployed neither fully regularly nor completely randomly. Analytical characterizations of the coverage properties of non-Poisson networks are generally difficult. To overcome this problem, we introduce the notion of the asymptotic deployment gain and show how it can be used to quantify the SINR distribution of a very general class of network models.