To investigate the spatial distribution and variability of carbon (C), nitrogen (N) and phosphorus (P) across large landscapes is important to address questions related to global warming, carbon sequestration, eutrophication, and cycling of biogeochemical properties. Understanding the spatially-explicit relationships between C, N and P and controlling environmental variables allows us to derive functional relationships and assess anthropogenic induced nutrient imbalances. A comprehensive landscape scale nutrient study was conducted in the subtropical Santa Fe River Watershed (SFRW) (3,585 km2) in north-central Florida. We used 141 soil samples collected at four depths across the watershed. Samples were collected following a random design, stratified by soil order and land use. Numerous ancillary environmental variables were derived for the watershed including a digital elevation model and derived topographic properties, Landsat ETM+ satellite imagery and derived indices and reflectance transformations, soil, climate, land use, and geology data. Soil samples were analyzed for total C, N, and Mehlich P and various physical and chemical fractions. Numerous parametric and non-parametric multivariate regression methods and hybrid geostatistical methods were used to relate soil properties to environmental variables. Besides identifying the prediction models that performed best we also investigated the strength and magnitude of environmental properties to predict a given target variable.