Abstract: Abstract: Contamination of suburban, agricultural soils with heavy metals draws great attention because of its potential threat to food safety and its detrimental effects on the ecosystem. The origins of soil heavy metals in the suburban interface are usually controlled by many factors, such as parent material, industrial activities, and agriculture. To decrease heavy metals pollution risks effectively in suburban areas and further to establish reliable protection measures, it is quite necessary to understand their sources and spatial patterns.The ordinary linear regression model (OLS) has been frequently used to analyze the relationship between soil heavy metals and their influential factors. However, OLS is only in a global or an average sense to estimate parameters, and it is unable to reflect spatial local variation or test spatial non-stationarity．Geographically weighted regression models (GWR) are a powerful tool for exploring spatial heterogeneity. The underlying idea of GWR is that parameters may be estimated anywhere in the study area given a dependent variable and a set of one or more independent variables which have been measured at known locations. Not only can it test spatial non-stationarity, but it can also provide the corresponding solutions. As a local model, GWR modeling has been applied in research on urban housing land prices and the spatial factors of economic development, but it has seldom been applied to the origins and spatial structure of soil heavy metals.A survey was conducted in this study to determine the possible sources of heavy metals in agricultural soils of the suburban area of Changsha. A total of 513 surface soil samples were collected, and the concentrations of Pb and Cd were analyzed. Typical influences on soil Pb and Cd concentration were identified from soil properties and geographic locations, such as soil pH, organic matter, alkali-hydro nitrogen, rapidly available phosphorus, rapidly available potassium, slowly available potassium, the distance from cropland to town, the distance from cropland to settlement, the distance from cropland to industrial construction sites, and the distance from cropland to a river. The OLS and GWR were applied to determine the relationships among both the influential factors and their spatial structure.The results indicate that spatial autocorrelations were detected for Pb and Cd. The high-high spatial clusters districts had high concentrations of Pb and Cd and were the most important regions for controlling the pollution risk of Pb and Cd in agricultural soil of the suburban area of Changsha. The GWR models for Pb and Cd had a better goodness-of-fit than OLS models and indicated the same tendency of spatial correlation between the Pb and Cd measured values with their estimated values. Soil Pb was highly significantly and positively related with Cd. The concentrations of soil pH, organic matter, alkali-hydro nitrogen and rapidly available phosphorus were the important influential factors for the content of Pb and Cd. The distance from cropland to river, from cropland to town, and from cropland to construction sites also had some influence on the concentrations of Pb and Cd in the agricultural soils of suburban Changsha.