Local populations within a species can become isolated by stochastic or adaptive processes, though it is most commonly the former that we quantify. Using presumably neutral markers we can assess the time-dependent process of genetic drift, and thereby quantify patterns of differentiation in support of the effective management of diversity. However, adaptive differences can be overlooked in these studies, and these are the very characteristics that we hope to conserve by managing neutral diversity. In this study, we used 16 hypothetically neutral microsatellite markers to investigate the genetic structure of the roundnose grenadier in the North Atlantic. We found that one locus was a clear outlier under directional selection, with FST values much greater than at the remaining loci. Differentiation between populations at this locus was related to depth, suggesting directional selection, presumably acting on a linked locus. Considering only the loci identified as neutral, there remained significant population structure over the region of the North Atlantic studied. In addition to a weak pattern of isolation by distance, we identified a putative barrier to gene flow between sample sites either side of the Charlie-Gibbs Fracture Zone, which marks the location where the sub-polar front crosses the Mid-Atlantic Ridge. This may reflect a boundary across which larvae are differentially distributed in separate current systems to some extent, promoting differentiation by drift. Structure due to both drift and apparent selection should be considered in management policy.