Fungal endophytes are suggested to manipulate host grass reproduction. Using different grass-endophyte combinations in a four year common garden experiment, we show that endophyte can retain control over the reproductive functions of the host. Plants were collected as seeds from two different environments, meadows and riverbanks. Natural endophyte infection (E+) increased reproductive effort of plants; 13 and 15 % higher proportion of E+ plants produced panicles compared to naturally uninfected (E-) and manipulatively endophyte-infected (ME+) plants, respectively. Meadow-origin E+ plants produced also higher number of panicles compared to E- in the last two years, and both meadow- and riverbank-origin ME+ plants produced higher number of panicles compared to ME- plants in most of the years. Pseudovivipary was recorded in 5 % of the plants each year. Pollen limitation appears not to induce pseudovivipary. Both E+ and ME+ plants produced higher number of seminiferous panicles compared to E- and ME- plants. The higher proportion of pseudoviviparism in ME+ plants compared to E+, E- and ME- plants from meadows suggests specific genotype-genotype cross-talk between endophyte and grass. Pseudovivipary was affected by year and primarily explained by plant reproductive effort since most vigorous plants produced more pseudoviviparous panicles. We propose that endophytes can promote host grass reproduction in a habitat-specific manner and depending on genetic compatibility between partners. However, reproductive behavior depended on maternal habitat and the annual environmental conditions highlighting the importance of long-term experiments to estimate the effects of endophytes on host plant ecology.