Yield mapping is one of the core tools of precision agriculture, showing the result of combined growing factors. In a series of yield maps collected along seasons it is possible to observe not only the spatial distribution of the productivity but also its spatial consistency among different seasons. This work proposes the study of distinct methods to analyze yield stability in grain crops regarding its potential for defining management zones from a historical sequence of yield maps. Two methods are herein used and applied into case studies to observe the existence of temporal consistency along harvest seasons. One method proposes the use of global filtering techniques for removal of defective yield data; averaging of the dense spatial data into grid-cells; and afterwards normalizing the cell-values towards their mean value. A second method proposes the use of local filtering to remove inconsistent neighboring data; attributing zone definitions to the point-yield-values by applying unsupervised fuzzy classification (Management Zone Analyst software); and clustering the data in cell-values retrieving: the average, median, and standard deviation of the cell-values, and the average and mode of the zone-classified points. The similarity among time-series cells was retrieved trough Spearman correlation for all the attributes obtained, along with their coefficient of variation. Until now, to the first method four case studies were submitted of selected field-plots, with yield data collected along four years, that were subset into smaller regions to safeguard genetic homogeneity and full yield data acquisition. A significant positive correlation was found among the cells for only two of these case studies, for which zones could also be visually observed. It is suggested that the reduced area of these subset-fields (down to 12 ha) reduces the possibility for existence of zones. To the second method two case studies were applied using full area yield data collected along 2 and 4 harvests. The filtering technique applied increased the spatial dependency among points by removing local inconsistency; the zone classification created from two to five zone classes visually observable in map scale; and the clustering of data into cells retrieved a total of 10 parameter classification attributes. Significant positive correlations, not lower than 0.568, were found among maps of corn yield for three year-seasons, reaching values of 0.752 among years of very distinct climatic conditions. Further studies have to be continually carried for comparison of same case studies to each method and exploring the combination of zone attributes; also parameters of cell-size clustering, removal of cells merged with high variance of points and intensity of pre-filtering of data still have their influence and sensitivity to be studied.