Physical processes in the field are greatly influenced by landscape attributes and, as a result, researchers expect to explain soil characteristics and crop performance using landscape properties. Typically, terrain attributes (e.g. slope curvature, slope gradient) are calculated from the digital elevation map (DEM) at the same (fixed) scale (lag). The scale of analysis influences the calculated value of terrain attributes and as a result the soil- and/or crop-landscape relationships. The objectives of this study were to determine: a) the scale at which individual landscape attributes should be calculated to maximize explanation of spatial variation in no-till corn (Zea mays L.) yield; and b) the best yield explanatory model, to include soil surface properties (texture, organic matter, bioavailable nutrients). We examined the relationship between yield and terrain attributes calculated at scales/lags ranging from 5 to 40 m in a 5-meter increment in four Kentucky (USA) fields. Using regression analysis, we observed that when landscape attributes were all calculated at a single lag, optimal explanatory regression of yield (R²= 0.19 to 0.54) did not occur at the smallest (5 m) lag, nor was it consistent across fields. In two fields, optimal explanatory regression for yield required attributes determined at different scales/lags. We conclude that landscape attributes calculated at different scale/DEM resolutions explain variability in corn yield better than the conventional “same lag” calculated attributes.