Precision agricultural technology promises to move crop production closer to a manufacturing paradigm, but analysis of yield monitor, sensor and other spatial data has proven difficult because correlation among neighboring observations often violates the assumptions of classical statistical analysis. When spatial structure is ignored variance estimates tend to be inflated and significance levels of test statistics are reduced. The gap between data analysis and site-specific recommendations has been identified as one of the key constraints on widespread adoption of precision agriculture technology. This paper compares four approaches that explicitly incorporate spatial correlation into regression models: (1) a spatial econometric approach; (2) a polynomial trend regression approach; (3) a classical nearest neighbor analysis; and (4) and a geostatistic approach. In the Argentine data studied, the spatial econometric, geostatistical approach and spatial trend analysis offered stronger statistical evidence of spatial heterogeniety of nitrogen response than the ordinary least squares or nearest neighbor analysis. All the spatial models led to the same economic conclusion, which is that variable rate nitrogen is potentially profitable. The spatial econometric analysis can be implemented on relatively small data sets that do not have enough observations for estimation of the semivariogram required by geostatistics. The spatial trend analysis can be implemented with ordinary least squares functions that are already available in some GIS software. In this study, the main benefit of using spatial regression analysis is increased confidence in the corn yield response estimates by management zone, and conclusions about the profitability of precision agriculture technologies.