The most adopted precision agriculture technique worldwide is the variable-rate fertilizer application based on soil grid sampling and followed by data interpolation to create soil fertility maps. However, most of the practitioners do not apply geostatistical analysis adequately on the data, creating maps through mathematical interpolators, like Inverse Distance Weight (IDW). Thus, just a minority of precision agriculture users performs geostatistical interpolation (kriging), while just a few of them make the analysis rigorously based on geostatistical assumptions. One of the effects that may influence on the quality of the maps is the spatial orientation effect (i.e. anisotropy and trend effect), that increase the ability in identifying specific behavior of the soil property on small scale. However, these effects are often not considered by the analyst. This might happen because the practitioner tends to expend several efforts and time performing geostatistical analysis, but frequently do not see high differences among the maps generated as well as do not know how to evaluate them. Thus, we tested different interpolation procedures (IDW, ordinary kriging without considering anisotropy and trend effect, and dealing with these both effects) in order to quantify fertilizer rate errors when performing proper geostatistical analysis (performing trend removal and anisotropy treatment) to build prescription maps. We used three sugarcane fields with high sampling density (more than one sample/ha) and evaluate potassium, phosphorus and lime prescription maps on specific validation sample-points. We identified that when the soil property shows strong spatially structured behavior (measured by the Moran´s Index), kriging considering trend and anisotropy effects yields better accuracy on the prescriptions maps. However, when the spatial distribution of the soil property is mostly randomized (low Moran´s Index and high nugget effect), IDW can be even better in quantifying the soil property and, as consequence, the fertilizer rate on specific points due to the kriging smoothing effect.