Initial studies of grass–endophyte mutualisms using Schedonorus arundinaceus cultivar Kentucky-31 infected with the vertically transmitted endophyte Epichloë coenophiala found strong, positive endophyte effects on host-grass invasion success. However, more recent work using different cultivars of S. arundinaceus has cast doubt on the ubiquity of this effect, at least as it pertains to S. arundinaceus–E. coenophiala. We investigated the generality of previous work on vertically transmitted Epichloë-associated grass invasiveness by studying a pair of very closely related species: S. pratensis and E. uncinata. Seven cultivars of S. pratensis and two cultivars of S. arundinaceus that were developed with high- or low-endophyte infection rate were broadcast seeded into 2 × 2-m plots in a tilled, old-field grassland community in a completely randomized block design. Schedonorus abundance, endophyte infection rate, and co-occurring vegetation were sampled 3, 4, 5, and 6 years after establishment, and the aboveground invertebrate community was sampled in S. pratensis plots 3 and 4 years after establishment. Endophyte infection did not enable the host grass to achieve high abundance in the plant community. Contrary to expectations, high-endophyte S. pratensis increased plant richness relative to low-endophyte cultivars. However, as expected, high-endophyte S. pratensis marginally decreased invertebrate taxon richness. Endophyte effects on vegetation and invertebrate community composition were inconsistent among cultivars and were weaker than temporal effects. The effect of the grass–Epichloë symbiosis on diversity is not generalizable, but rather specific to species, cultivar, infection, and potentially site. Examining grass–endophyte systems using multiple cultivars and species replicated among sites will be important to determine the range of conditions in which endophyte associations benefit host grass performance and have subsequent effects on co-occurring biotic communities.