The understanding of ecophysiological basis of wheat (Triticum aestivum L.) grain yield potential provides a useful framework to complement conventional breeding aimed at achieving genetic gains. This study analyzed the ecophysiological performance of an elite wheat mapping population (105 double-haploid lines derived from two modern cultivars, Bacanora and Weebil, with similar phenology but different and stable combinations of grain number per area unit (GN) and grain weight (GW) resulting in high grain yield) grown in four contrasting high-yielding environments, to determine the most successful strategies to increase grain yield potential. Main effect of environment on grain yield was significant (p < 0.0001) but the genotypic component was larger than genotype × environment interaction (30%). A robust and positive relationship between grain yield and biomass production was observed across all environments (r2 > 0.82, p < 0.0001), and relatively high harvest indexes were expressed (0.39–0.51). While GN was clearly the dominant numerical component in terms of association with grain yield (r2 > 0.51, p < 0.0001), a wide range in both components (i.e., GN and GW) was observed across all environments. This population represents a valuable resource for prebreeding studies, as the transgressive segregation in physiological and numerical yield components in combination with favorable expression of all agronomic traits could allow a fine phenotyping and mapping to identify key traits and quantitative trait loci linked with grain yield.