Switching off base stations (BSs) when the activity in the cell is relatively low, often referred to as cell sleeping, is one possible method of reducing energy consumption in macrocellular networks. However, this method may as well reduce the coverage and hence it is not evident whether the energy savings can compensate for the network throughput reduction. In this paper, we propose and analyze two sleep mode strategies, namely random and strategic sleeping, for energy efficient cellular networks. Using stochastic geometry, we model the switching off BS operation and obtain analytical results for the coverage probabilities and the area spectral efficiency under different cell sleeping strategies. Specifically, the effect of proposed sleep mode policies on the power consumption and on the energy efficiency is investigated. We first show that the performance gains depend on the level of background noise. Furthermore, numerical results show the effectiveness of an activity-based sleeping strategy in maximizing the energy efficiency of macrocellular networks.