The interaction of meddies with a complex distribution of seamounts is studied in a three-layer quasi-geostrophic model on the f-plane. This study aims at understanding if and how this seamount chain can represent a barrier to the propagation of these eddies and how it can be involved in their decay. The eddies are idealized as vortex patches in the middle layer, interacting with a regional cyclonic current and with ten idealized seamounts. The numerical code is based on the contour surgery technique. The initial position, radius, shape, number and polarity of the eddies are varied. The main results are the following: (1) Though they do not describe the unsteady flow, the streamlines of the regional and topographic flow provide a useful estimate of the vortex trajectories, in particular towards the major seamounts, where stronger velocity shears are expected. (2) The tallest and widest seamounts which have the largest vorticity reservoir are able to considerably erode the vortices, but also to draw anticyclones towards the seamount top. The ability of narrower seamounts to erode vortices is related to their multiplicity. (3) Only 1/3 of the anticyclones with about 30-km radius reach the southern boundary of the seamount chain, and their erosion is larger than 50 %. The other anticyclones are either completely eroded or trapped over a wide seamount top. Cyclones are less affected by seamounts because they oppose the topographic draft towards the seamount top and they drift along the side of the seamount. (4) Large vortices resist topographic erosion more efficiently. The rate of erosion grows from a few percent to about 35-50 % as the vortex radius decreases from about 60 to 30 km. Small cyclones are not eroded, contrary to small anticyclones (which completely decay), in relation with the different trajectories of these eddies in the vicinity of the seamounts. (5) The detailed vortex shape does not appear critical for their evolution, if they are close enough to the seamount chain initially. The interaction between a group of vortices initially north of the seamount chain can modify their trajectory to such an extent that they finally avoid collision with seamounts. (6) Finally, meddy trajectories across the Horseshoe Seamounts (data from the AMUSE experiment) show qualitative similarity with the vortex paths in the model. Several events of vortex decay also occur at comparable locations (in particular over the wide and tall seamounts) in the model and observations.