Most discussions of the digital divide treat it as a "North-South" issue, but the conventional dichotomy doesn't apply to cell phones in Sub-Saharan Africa. Although almost all Sub-Saharan countries are poor by international standards, they exhibit great disparities in coverage by cell telephone systems. Buys, Dasgupta, Thomas and Wheeler investigate the determinants of these disparities with a spatially-disaggregated model that employs locational information for cell-phone towers across over 990,000 4.6-km grid squares in Sub-Saharan Africa. Using probit techniques, a probability model with adjustments for spatial autocorrelation has been estimated that relates the likelihood of cell-tower location within a grid square to potential market size (proximate population); installation and maintenance cost factors related to accessibility (elevation, slope, distance from a main road, distance from the nearest large city); and national competition policy. Probit estimates indicate strong, significant results for the supply-demand variables, and very strong results for the competition policy index. Simulations based on the econometric results suggest that a generalized improvement in competition policy to a level that currently characterizes the best-performing states in Sub-Saharan Africa could lead to huge improvements in cell-phone area coverage for many states currently with poor policy performance, and an overall coverage increase of nearly 100 percent.