Odense Fjord (Denmark) has suffered from a drastic decline in eelgrass (Zostera marina) coverage during the last decades. In 1983 eelgrass still covered about 25% of the estuary, which in 2005 was reduced to less than 2%. The alarming decline in the past decades initiated preliminary restoration activities, where it was questioned whether the present low eelgrass biomass is able to produce a sustainable seed bank to support natural recovery. Field studies verified that the seed bank was hampered. Laboratory experiments were conducted to determine 1) seed dispersion along the sediment surface and in the water column; 2) settling rates of seeds and flowering shoots; 3) critical current speed for seed movement; 4) floating dynamic of flowering shoots and 5) seed dropping dynamics during transport of rafting shoots. These parameters supported the development of an agent based model (ABM) predicting seed movements in estuaries. The model handled two ways of seed dispersal: 1) seeds dropped in eelgrass beds and transport by hydrodynamic forces along the seabed 2) seeds released by rafting shoots. This setup allowed assessment of both eelgrass seed loss and potential connectivity between beds. Seed losses were divided into direct losses, such as seeds lost on land due to desiccation or external boundary, and indirect losses affecting seedling establishment. The model estimates that app. 92% of the seeds would be retained in the Odense fjord, while only 5.0% of the seeds ended up in areas supporting seedling establishment. Eelgrass seeds were also found in areas with insufficient light, high hydrodynamic pressure, excessive sediment reworking by lugworms or poor anchoring capacity. In addition, the model showed potential bed connectivity via rafting shoots, but also with individual seed movement along the bottom, when beds were not separated by deep areas, such as boating channels.