Abstract. Information on the relationships between quantitative traits affecting yields is extremely important for winter wheat. For it, the annual genotype*environment interaction is palpable and often masks the influence of individual traits on grain yield. The aim of the study is to determine the traits through the selection of which the grain yield could be significantly increased in the future. The data from three field multifactorial experiments were used (FERT, PGR, ABC), in which a significant influence of various factors (year, point, density, fertilization) on the size and variation of all studied traits was established. In the database thus formed the observed strong variance in the values of the traits is a great prerequisite for the established correlations to be accepted with a high degree of reliability. The mutual influence in the formation of each of the traits is a good basis for their grouping, according to the type of their effect on yield. 1) The characteristics, number of grains per m2 (NGm), grain weight per spike (WGS) and number of productive tillers per m2 (NPT) have a significantly positive effect on grain yield, 2) the weight per 1000 grains (TGW) and number of grains in spike (NGS) are traits that have a direct effect, but it is unstable in manifestation and 3) the traits, as height of stem (HOS), total plant biomass (TBM), and harvest index (HI) do not show a direct effect on grain yields. A significant increase in yield in the breeding of winter wheat can be achieved by increasing the number of grains per unit area (NGm). This is possible while maintaining the achieved level of number of grains in spike (NGS) with a parallel increase of tillering productive ability (NPT). The increase of this trait by selection should be taken into account when reducing the grain size (TGW). This will increase the chance of increasing the number of grains in the spike (NGS), will reduce the weight of the grain per spike (WGS), which in turn will be a prerequisite for optimizing the stability of lodging