Vineyard variability within the fields is well known by grape growers, producing different plant responses and fruit characteristics. Many technologies have been developed in last recent decades in order to assess this spatial variability, including remote sensing and soil sensors. In this paper we study the possibility of creating a stable classification system that better provides useful information for the grower, especially in terms of grape batch quality sorting. The work was carried out during 4 years in a rain-fed Tempranillo vineyard located in Rioja (Spain). NDVI was extracted from airborne imagery, and soil conductivity (EC) data was acquired by an EM38 sensor. Fifty-four vines were sampled at véraison for vegetative parameters and before harvest for yield and grape analysis. An Isocluster unsupervised classification in two classes was performed in 5 different ways, combining NDVI maps individually, collectively and combined with EC. The target vines were assigned in different zones depending on the clustering combination. Analysis of variance was performed in order to verify the ability of the combinations to provide the most accurate information. All combinations showed a similar behaviour concerning vegetative parameters. Yield parameters classify better by the EC-based clustering, whilst maturity grape parameters seemed to give more accuracy by combining all NDVIs and EC. Quality grape parameters (anthocyanins and phenolics), presented similar results for all combinations except for the NDVI map of the individual year, where the results were poorer. This results reveal that stable parameters (EC or/and NDVI all-together) clustering outcomes in better information for a vineyard zonal management strategy.