Recruitment is a crucial event in the plant life cycle that is very sensitive to interaction with established vegetation. Based on a large comparative experiment, we tested the hypothesis that the components of recruitment--emergence time and rate, seedling survival and biomass--differ in response to plant-plant interactions during recruitment. The consequences for the population are predicted with a simple demographic model assessing the response of seed production. In a common garden experiment, we recorded the recruitment of four target species in an individual-based survey protocol. A total of 7,680 seeds were sown within 20 neighbourhoods, consisting of 19 mono-specific herbaceous stands and a control treatment without vegetation. We measured transmitted light, temperature and moisture at soil surface to characterise the environmental conditions within neighbourhoods. The mean height of neighbours controlled temperature buffering and light interception and thus depicted the interaction gradient. Emergence rate and time increased with neighbour height in two of the four target species, while seedling survival and biomass significantly decreased with neighbour height in three and all four target species, respectively. We recorded a shift in seedling neighbour interactions under the tallest neighbours that largely favoured emergence but strongly depressed seedling survival and biomass. The components of recruitment were predicted to differ in their impact on later adult performance. Biomass strongly contributed to predicted seed production in three target species, and emergence had an equal or greater impact on a fourth species. These results confirm the fundamental role of plant-plant interactions in the recruitment of herbaceous species through a complex combination of habitat amelioration, which facilitates emergence and light competition, which in turn limits seedling survival and biomass.