Many studies have shown effects of plants species on fungal communities, but these are often confounded with soil effects. Thus, the specific role of plant species in structuring rhizospheric and soil fungal communities is poorly described. Our study used microcosms in which plants were grown under artificial conditions to bridge this gap. Two perennial grasses dominating subalpine grasslands, Festuca paniculata and Dactylis glomerata, were grown at two levels of fertilization on standard soil. Fungal communities were determined by 454 pyrosequencing of the internal transcribed spacer 1 region. Among the fungal communities characterized by the primers used, original communities were associated to each plant species and also diverged between rhizosphere and bulk soils within each plant species, though there were no significant fertilization effects. Differences regarded global composition of the fungal communities and abundant molecular operational taxonomic units (MOTUs). Both plant species and location effects were reflected more in the abundance than in the composition of MOTUs. The observed differences in fungal communities coincide with differing strategies of plant root growth, with D. glomerata having greater root mass, length, and area than F. paniculata. Our study, by dissociating soil effects from plant effects, demonstrated that plant species exert a key control on soil fungi. We suggest that such effects may be linked to inter-specific differences in root traits and their consequences on nitrogen uptake.