In recent decades, increases in wheat yield have been achieved mainly through increases in grain number per m2 (GNM2) rather than through increases in average grain weight (AGW). Using AGW as a lever to increase yield would require avoidance of the negative relationship between GNM2 and AGW. It is usually supposed that this trade-off arises from an increase in the proportion of small grains as GNM2 rises. The proportional increase in small grains being the result either of (1) an increase in the proportion of secondary tillers in the spike population or (2) of an increase in the proportion of grains located in distal positions within each spike. Either or both of these two populational effects would tend to mask any true genotypic differences in AGW. The existence of these constitutive differences has already been proposed, but without considering the full extent of the populational confounding effects. Identification of a component of the constitutive genetic determinism of AGW - one that is truly independent of GNM2 - could contribute to cultivar developments that would lead to further increases in grain yield under future target environments. To address this question, we analysed populational effects on AGW in four, modern, well-adapted bread-wheat cultivars. The four chosen cultivars show similar grain yields but contrasting AGWs. The analysis of populational effects was carried out at three hierarchical levels (the plot, the spike and the single grain) and under two contrasting environmental conditions (well-watered vs waterdeficit conditions). Regardless of the environment, no (or only slight) differences in individual spike size were observed between cultivars. Furthermore the weak relationship between spike size and AGWdemonstrates that AGW differences between cultivars cannot be attributed to spike-level populational effects. Meanwhile, the analysis of individual grain mass distributions, showed that the differences in AGW between cultivars, originated from shifts in the whole grain-mass distribution, rather than from shape changes in the grain-mass distribution. This clearly indicates that AGW differences between cultivars cannot be attributed to populational effects at the individual grain level. The analysis carried out at both spike and individual grain levels indicates that the AGW differences between cultivars are largely constitutive, so that increases in grain yield through AGW may be considered independently of the GNM2 : AGW trade-off. Taken together, these findings offer a new perspective for the genetic improvement of wheat, and one that should lead to further increases in yield.