Abstract Pastures that include annual and perennial grasses can play a strategic role in increasing the sustainability of temperate livestock systems. However, the impacts of the competition between annual and perennial genetically improved forage species on pastures has scarcely been studied. We evaluated the early dynamics of tillering, carbon partitioning and both intra- and interspecific competition in – or between – an annual (Bromus willdenowii; prairie grass; Bw) and a perennial (Dactylis glomerata; orchard grass; Dg) C3 grass species. We sowed 24 0.18 m2 microplots in May 2008. Each microplot constituted an experimental unit and contained five sowing lines spaced 0.175 m apart. On three of the lines, we sowed different combinations of the mentioned grasses, while on the two lines between them, we sowed red clover (Trifolium pratense). We randomly applied a factorial array of eight treatments—2 species (Dg vs. Bw) × 2 types of competition (intra vs. interspecific) × 2 levels of winter nitrogen fertilization (0 vs. 170 kg of N ha-1)—designed in three complete random blocks. Fertilization did not have any significant effect on any species. The evaluation period lasted 109 days after sowing (das), separated into three phases (0-53 das, 59-90 das and 90-109 das). The variables recorded were aerial biomass (leaves, sheets and senescent material), root biomass and tiller density per species. As expected, annual Bw inhibited the growth of perennial Dg at the latest stage of establishment but surprisingly promoted it at the earliest stage. This work suggested that a low initial density of an annual species should optimize the initial facilitating effect of the annual species on the perennial species with its subsequent inhibitory effect.