Symbiotic associations between grasses and vertically transmitted endophytic fungi are widespread in nature. Within grass populations, changes in the frequency of infected plants are driven by influence of the endophyte on the fitness of their hosts and by the efficiency of endophyte transmission from parent plants to their offspring. During the seed stage, the endophyte might influence the fitness of its host by affecting the rate of seed viability loss, whereas the efficiency of endophyte transmission is affected by losses of viability of the fungus within viable seeds. We assessed the viability losses of Lolium multiflorum seeds with high and low level of infection of the endophyte Neotyphodium occultans, as well as the loss of viability of the fungus itself, under accelerated seed ageing and under field conditions. Starting with high endophyte-infected accessions of L. multiflorum, we produced their low endophyte-infected counterparts by treating seeds with a fungicide, and subsequently multiplying seeds in adjacent plots allowing pollen exchange. In our accelerated ageing experiments, which included three accessions, high endophyte-infected seeds lost viability significantly faster than their low endophyte-infected counterpart, for only one accession. High endophyte-infected seeds of this particular accession absorbed more water than low endophyte-infected seeds. In contrast, the endophyte lost viability within live seeds of all three accessions, as the proportions of viable seeds producing infected seedlings decreased over time. In our field experiment, which included only one accession, high endophyte-infected seed lost viability significantly but only slightly faster than low endophyte-infected seed. In contrast, the loss of viability of the endophyte was substantial as the proportions of viable seeds producing infected seedlings decreased greatly over time. Moving the seeds from the air to the soil surface (simulating seed dispersion off the spikes) decreased substantially the rate of seed viability loss, but increased the rate of endophyte viability loss. Our experiments suggest that, in ageing seed pools, endophyte viability loss and differential seed mortality determine decreases in the proportions of endophyte-infected seeds in L. multiflorum. Endophyte viability loss within live seeds contributes substantially more to infection frequency changes than differential viability losses of infected and non-infected seeds.