Modeling the spatial distribution of base stations (BSs) is essential for evaluating different network performance metrics in cellular networks. In this paper, we consider the spatial distribution data of actual BS deployments in Tier-1 cities of India for one of the leading network providers. To model this data, we consider the widely used homogeneous Poisson Point Process as the benchmark. We evaluate the performance of the Poisson distribution along with Discrete Exponential, Discrete Weibull, and Zipf-Mandelbrot distributions for modeling the distribution of number of BSs in a given area. Further, we model the spatial location of BSs using Uniform, Gaussian, and Laplace distributions. We compare the performance of the various distributions using Goodness of Fit tests. The numerical results indicate that the location of BSs can be accurately modeled as a Gaussian distribution, while, the number of BSs in the Indian scenario is best modeled as a Discrete Exponential distribution. © 2017 IEEE.