We consider the impact on pilot contamination of reducing cell size in massive MIMO cellular networks. In the limit of a large number of antennas, pilot contamination dictates the capacity limit of such networks. In this work, we examine how capacity of massive MIMO cellular networks changes as a function of cell size. In particular, we introduce the concept of copilot distance and formulate the worst-case capacity of a typical user in the network as a function of this distance. This formulation captures the effect upon capacity of all system parameters such as cell radius, user density and the fraction of time devoted to channel estimation. Based on this formulation, we show how cell size reduction mitigates the pilot contamination and hence increases the capacity of each user as well as the number of supported users. We also provide numerical and simulation results to confirm our analysis.