This article outlines a new interpretation of the changing geography of North American cities and suburbs from about 1900 to 1950. By North American,we refer to Canada and the United States because, as we suggest in section 1, the processes and patterns of urban development in these two countries were broadly similar. We recognize the importance of jobs, especially in manufacturing, in shaping the geography of metropolitan areas (section 2). We see patterns of residence as being dependent on job location and transportation but also on variations in how homes were used as sites of unpaid work (section 3) and in the ways that land was developed (section 4). Finally, we see the political geography of metropolitan areas as both an expression and a determinant of social patterns (section 5). Using this scaffolding, we build our argument. We note that decentralization of employment was well under way by 1900, typically in polycentric clusters. It involved offices and stores, as well as factories, and encouraged the large-scale suburbanization of workers and immigrants. As a result, although individually homogeneous, suburbs in the aggregate were socially diverse, as were most central cities. We believe that it is inaccurate to think of cities in this perios as poor and suburbs as affluent. Perhaps the most original and controversial element to our argument is the one that we present last. Contradicting the idea that suburban self-rule guaranteed the social exclusivity of the suburbs, we argue that the fragmentation of municipal government allowed all sorts of people to settle at the fringe. It blurred rather than sharpened the line between city and suburb, rendering such a distinction moot.