Chinook salmon (Oncorhynchus tshawytscha) are widely distributed across the globe, with native stocks in the North Pacific Ocean and self-sustained populations in both the Northern and Southern hemispheres. In their native range, Chinook salmon face many conservation and management challenges, including depleted stocks, loss of genetic diversity, and hatchery influences, whereas naturalized range expansion poses a threat to novel ecosystems. Therefore, ways to improve stock discrimination would be a useful tool for fishery managers. Here, we evaluated otolith shape variation in Chinook salmon as a potential tool for stock discrimination using wavelet coefficients and Fourier harmonics in three case studies at multiple spatial scales. We adopted a simple Classification Tree model that used otolith shape variation to separate Chinook salmon groups. We found best performance of the model occurring between hemispheres, followed by Oregon basins, within-watershed Elk River, Oregon, and lastly among South American basins. Otolith shape analysis is a promising tool for stock discrimination if used in conjunction with other methods to better understand plasticity of anadromous species that use pan-environmental systems.