Geographically weighted regression (GWR) analysis and on-farm precision experiments (OFPE) allow the quantification of the within-field spatial variability of crop response to controllable inputs. Few studies have quantified the magnitude and consistency of these responses across years, and their relationship with site-specific characteristics. In this study we: (1) assessed the within-field spatial variability of corn response to nitrogen (NR) and seed (SR) rates using OFPE data and GWR models; (2) determined the spatial agreement between NR and SR response classes, and their temporal stability; and (3) modeled the effect of weather and site-specific features on the spatial distribution of the responses. Response maps were estimated by fitting GWR models to yields and as-applied NR and SR from 14 OFPE. Response maps classified into positive and non-positive response classes were spatially joined to weather, soil, and landscape covariates to train a random forest model. Zones with positive responses to NR and SR were detected in all fields. In most cases, these zones were inconsistent across years due to weather. The higher spatial agreement between NR and SR response classes for the same year suggests that factors controlling the crop response to these inputs are similar. In both cases, weather variables were the most important predictors followed by landscape and soil attributes. Whereas weather variables acted at a field level, i.e. field-year, the latter reflected the effect of within-field variability on crop response. To reduce this seasonal uncertainty, future research should consider conducting larger numbers of OFPE across multiple seasons.