Scoring the structure quality of the soil with the Visual Evaluation of the Soil Structure (VESS) spade test receives growing interest due to its simplicity, reliability and the quality of the evaluation provided. Sampling requirements to achieve a predefined quality of estimation, however, were not defined. This paper aims at filling this knowledge gap. The spatial variability of VESS scoring is analysed in two field, one with homogeneous micro-topography and vegetation growth and one showing spots of two distinct states with respect to these criteria. The semi-variograms showed purely random distribution of the scores from 6–10 to > 150 m of inter distances between observations, thus allowing to calculate the number of spade tests required to reach objectives of accuracy or minimum detectable difference. The two zones of the heterogeneous field, and the homogeneous field, showed the same coeffcient of variation of 11%, which is small. Therefore, 5 spade tests only are enough to detect a 0.5 change of the scoring. This number is smaller than usually recommended or performed. The classical recommendation to sample in homogeneous vegetation growth is supported by our results. These results probably apply to most situations and can be used to design monitoring protocols.