We investigated the interactive effects of soil phosphorus (P) heterogeneity, plant density and mycorrhizal symbiosis on plant growth and size variability of Trifolium subterraneum. We set up mesocosms (trays 49×49 cm and 12 cm deep) with the same amount of available P, but distributed either homogeneously or heterogeneously, in randomly arranged cells (7×7 cm each) with high or low available P. The trays were planted with either 1 or 4 seedlings of T. subterraneum per cell. Half of the trays were inoculated with spores of the mycorrhizal fungus Gigaspora margarita. We harvested the plants when leaves just started to overlap, 8 weeks after planting. Plants growing in high P cells had the lowest percentage infection, but the highest mean shoot and root biomass and root length. The mean size of the plants in each cell was determined mainly by local P concentration. However, in plants growing in high density, low P cells, ca. 20% of the variability in plant biomass was explained by the number of adjacent cells with high P. Patchy trays had the highest total shoot biomass, independently of mycorrhizal infection or plant density. Inoculated trays (M) had higher total shoot biomass and relative competition intensity (measured as reduction in plant biomass due to increased density) than non-inoculated trays (NM). Plant density reduced the plant response to mycorrhizal infection, and its effect was independent of P distribution. All populations growing in patchy trays, and low density mycorrhizal ones, had the highest plant-size inequality, presumably because patchy distribution of P and mycorrhizal infection increased competitive asymmetry. We conclude that mycorrhizal symbiosis has the potential to strongly influence plant population structure when soil nutrient distribution is heterogeneous because it promotes pre-emption of limiting resources.