Concrete quality is influenced to a high degree by the granulometry of the aggregates used. In the construction industry, the grain size coefficient of K‐coefficient is used as a quantity for the characterization of granulometry. In practice, this quantity is determined experimentally, and it is therefore a random variable influenced by material heterogeneity and errors in sampling, sample preparation and size analysis. Depending on the final use, in concrete production often an aggregate is required, the grain size coefficient of which can be adjusted to a given value Krequ with a prescribed precision. For this reason, continuous supervision of the aggregate quality is necessary using, e. g., automatic sampling and opto‐electronic measuring equipment. In order to determine the optimum conditions for the installation of such equipment, the variances of the possible errors must be established. Further, methods for their restriction are necessary. In this paper, investigations involving hand sampling and hand sieving and on the example of a reference gravel plant are reported. The investigations reveal that, for the materials involved, primary samples of at least 1600 g were necessary, which then could be reduced to 200 g by sample splitting, in order to obtain the accuracy desired.