Almost all natural plant communities contain arbuscular mycorrhizal fungi (AMF). We hypothesized that the species composition of AMF communities could have the potential to determine plant community structure if the growth response to different AMF species or to communities of AMF species varies among plant species. To test the existence of such a differential response we conducted a pot experiment where each of three plant species Hieracium pilosella Bromus erectus and Festuca ovina were inoculated with each of four AMF species or with a mixture of these four AMF species or were uninoculated. The AMF species originated from a calcareous grassland in which the three plant species also coexisted. We obtained three pieces of evidence suggesting that AMF have the potential to determine plant community structure. First plant species differed in their dependency on AMF thus varying in degree of benefit received. Second specific AMF species and a mixture of these AMF species had significantly different effects on several plant growth variables and these effects were not the same on each plant species. Third the amount of variation in the growth response of a plant species to four AMF species and to the mixture of AMF species differed among the plant species. Hieracium differed greatly in its growth response to several AMF species while Bromus did not exhibit much variation in its response to different AMF species. The varying mycorrhizal dependency of different plant species has previously been proposed as a mechanism determining plant community structure. However we found that the mycorrhizal dependency of a plant species can vary greatly because of differential growth responses to specific AMF species compared to the growth of the uninoculated plants. Consequently mycorrhizal dependency as a measure indicating how much a plant depends on AMF for its growth is not necessarily a fixed value and therefore cannot be used as a definitive term. In addition those plant species with highly variable responses to single AMF species or to combinations of AMF species (AMF communities) will be strongly affected by the specific species of AMF that occupy their roots in contrast to plant species that do not respond differently to different AMF species. We conclude that through their differential effects on plant growth AMF species that co-occur as natural AMF communities have the potential to determine plant community structure and that future studies on plant population and community structure need to consider the strength of their role as a determinant.