Uncertainty associated to analytical results is an issue of major interest for the whole analytical community. A large effort has been made to improve analytical techniques and procedures aimed to achieve a well characterized uncertainty associated with analysis. However, it is becoming increasingly recognised that uncertainty deriving from sampling and subsampling can even dominate the global uncertainty budget. A study on subsampling activities on different soil typologies has been performed by granulometry determinations. The differences between sieving methodologies based on both wet and dry mode have been studied. Subsampling is approached by replicated measurements providing a quantitative assessment of the distribution heterogeneity, a suitable method validation scheme and an empirical determination of uncertainty.