Land-use change is one of the most important drivers of soil change. Soil changes that follow agricultural and forest management have been well-studied in throughout the world, but the impacts of urban land use on soils have received little attention. Indeed, most knowledge of urban soils is derived from an engineering, rather than environmental perspective. Urban soil science is complex because of the diverse ways that people in cities manage soils (e.g. leveling, fertilization, paving, landscaping), but also because urban soils often have a diverse history of pre-urban land use. Here we present research that reveals the impact of pre-urban land use (agriculture versus native Sonoran Desert) on the soils beneath residential yards of Phoenix, AZ, USA. We compared surface soil (0 to 10 cm depth) organic matter (OM), carbon (C), nitrogen (N), inorganic P (Pav), and soluble ion (EC) concentrations in residential yards converted from farms (n = 16) to concentrations in yards developed on native desert (n = 23). Our goal was to determine whether a legacy of past agricultural use is detectable in the soil chemistry of contemporary residential lawns. Pools of OM, C, N, and EC were twice as great in yards that were previously farmed as in yards that were developed from desert, and the pools remained elevated 40 y after urbanization. Nitrogen accumulation rates (1.5 g m-2 y-1) in residential soils were not affected by prior land use, suggesting that home owners do not adjust rates of residential fertilizer application to account for the high fertility of previously farmed soils. Bioavailable, inorganic phosphorus (Pav) was elevated in soil with a recent agrarian past, but this signal disappeared after 10-30 y of residential use because Pav accumulated in yards developed on native desert soils. We compared our results from urban soils to an expansive (n = 68) set of native desert soils to confirm that patterns we observed in the city do not result from preferential cultivation and urbanization of fertile sites. Our results suggest a “direct agrarian legacy,” wherein agricultural impacts on soil chemistry endure urbanization, more so than an “indirect legacy,” wherein contemporary land management is shaped by prior land use. Unlike more humid regions, agriculture in dry lands may sequester material in soils, where it remains despite urbanization. These results have important implications for understanding feedbacks between soils and global change, as both agricultural development and urbanization are expected to proceed more rapidly in arid regions than temperate regions in the coming decades.