It is increasingly recognized that belowground responses to vegetation change are closely linked to plant functional traits. However, our understanding is limited concerning the relative importance of different plant traits for soil functions and of the mechanisms by which traits influence soil properties in the real world. Here we test the hypothesis that taller species, or those with complex rooting structures, are associated with high rates of nutrient and carbon (C) cycling in grassland. We further hypothesized that communities dominated by species with deeper roots may be more resilient to drought. These hypotheses were tested in a 3‐yr grassland restoration experiment on degraded ex‐arable land in southern England. We sowed three trait‐based plant functional groups, assembled using database derived values of plant traits, and their combinations into bare soil. This formed a range of plant trait syndromes onto which we superimposed a simulated drought 2 yr after initial establishment. We found strong evidence that community weighted mean (CWM) of plant height is negatively associated with soil nitrogen cycling and availability and soil multifunctionality. We propose that this was due to an exploitative resource capture strategy that was inappropriate in shallow chalk soils. Further, complexity of root architecture was positively related to soil multifunctionality throughout the season, with fine fibrous roots being associated with greater rates of nutrient cycling. Drought resistance of soil functions including ecosystem respiration, mineralization, and nitrification were positively related to functional divergence of rooting depth, indicating that, in shallow chalk soils, a range of water capture strategies is necessary to maintain functions. Finally, after 3 yr of the experiment, we did not detect any links between the plant traits and microbial communities, supporting the finding that traits based on plant structure and resource foraging capacity are the main variables driving soil function in the early years of grassland conversion. We suggest that screening recently restored grassland communities for potential soil multifunctionality and drought resilience may be possible based on rooting architecture and plant height. These results indicate that informed assembly of plant communities based on plant traits could aid in the restoration of functioning in degraded soil.