Determining spatio-temporal distributions of fish populations is of interest to marine ecology, in general, and to fisheries science in particular. Genetic mixed-stock analysis is routinely applied in several anadromous fishes for determining migratory routes and timing but has rarely been used for marine fishes, for which population differentiation is commonly weak and the method presumably less powerful. We used microsatellite information for Northeast Atlantic herring Clupea harengus L. populations and mixed stocks to address 2 questions. We used simulated mixture samples and 3 different statistical approaches to determine whether mixed stock composition could be determined with accuracy. Simulations showed that the applied approaches and mixture samples of 100 individuals enabled detailed composition analyses on a regional level, with resolution for tracing the ecologically dominant Rügen (Greifswalder Bodden) herring population. We then estimated spatio-temporal variation in herring migratory behaviour in the Skagerrak from 17 mixed samples collected over 2 seasons and 2 yr, and identified hitherto undescribed differences in distributions among populations that feed and winter in the area.