The ability to predict animal movement based on environmental change is essential for understanding the dynamic nature of their spatial ecology, and in turn the effectiveness of conservation strategies. We used a large marine predator that displays partial migration (the tiger shark Galeocerdo cuvier) as a model to test the role of oceanic conditions in predicting the space-use of different size classes. By using generalized additive mixed models (GAMMs), we revealed that environmental variables (sea surface temperature, primary productivity, thermal fronts, and bathymetry) had much greater predictive power for the movements of large, migratory tiger sharks than for small, resident individuals. We also found that coverage of tiger shark movements within “shark sanctuaries” (protected areas specifically for sharks) in the northwest Atlantic could be increased from 12 to 52% through inclusion of Bermuda’s waters. However, as large tiger sharks are migratory, over 80% of potential longline fisheries interactions would still occur outside the boundaries of even the expanded protected areas. This emphasises that management of highly migratory species needs to be dynamic and account for changing interactions with fisheries over time, which in a changing climate may rely on predicting movements based on oceanic conditions to be effective.