Ignoring distortions, the social benefit from an urban transportation improvement is typically measured as the reduction in transportation costs at a particular location, holding travel fixed at the pre-improvement level, summed over all locations. The result has been seen in specific urban transportation models, however, not formally derived as a general result. Through a transformation of variables first employed in Arnott and Stiglitz (1981), this paper adopts a di↵erent perspective to generalize the result. From this perspective, the benefits of a transportation improvement derive from the increase in the residential land area of better accessibility that the improvement brings about. The generalized result tells that a transportation improvement changes the land area at varying levels of accessibility, as well as land rents throughout the city, however, the social benefit of the transportation improvement is then measured as the increase in aggregate di↵erential land rents due to a change in land area at each level of accessibility, rather than that due to a change in land rent, by holding the function relating land rent to accessibility at its pre-improvement levels. This paper presents this result by using a basic monocentric model as a vehicle of demonstration and then shows that the result generalizes beyond the geographic features of the city, to a broad class of first-best urban economies with multiple transportation modes and employment centers and multiple household groups.