Our hypothesis was that simple models can be applied to predict yield by using only those yield data which spatially coincide with the soil data and the remaining yield data and the models can be used to test different sampling and interpolation approaches commonly applied in precision agriculture and to better predict soil variables at not observed locations. Three strategies for composite sample collection were compared in our study. Point samples were taken 1.) along lines within homogenous NDVI areas, 2.) along lines within homogenous electric conductivity scan zones and 3.) in circles around predefined regular grid points. Multiple regression models were developed to predict yields. Digital elevation, five to eight soil variables and in one case three agrotechnical variables (variable rate fertilizer use and seeding) were retained in the prediction equations with R2 values of 0.557, 0.248 and 0.191 for circular, soil EC based and NDVI based sampling, respectively. Spline interpolation method proved to be the best in two cases and IDW method was the best in the third case. The attempt to predict soil variables with fine spatial detail brought mixed results.