Given the increasingly important role of the Internet in education, healthcare, and other essential services, it is important that we develop an understanding of the "digital divide." Despite the widespread diffusion of the Web and related technologies, pockets remain where the Internet is used sparingly, if at all. There are large geographic variations, as well as variations across ethnic and racial lines. Prior research suggests that individual, household, and regional differences are responsible for this disparity. We argue for an alternative explanation: Individual choice is subject to social influence ("peer effects") that emanates from geographic proximity; this influence is the cause of the excess variation. We test this assertion with empirical analysis of a data set compiled from a number of sources. We find, first, that widespread Internet use among people who live in proximity has a direct effect on an individual's propensity to go online. Using data on residential segregation, we test the proposition that the Internet usage patterns of people who live in more ethnically isolated regions will more closely resemble usage patterns of their ethnic group. Finally, we examine the moderating impact of housing density and directly measured social interactions on the relationship between Internet use and peer effects. Results are consistent across analyses and provide strong evidence of peer effects, suggesting that individual Internet use is influenced by local patterns of usage. Implications for public policy and the diffusion of the Internet are discussed.