ABSTRACTMuch effort has been devoted to predicting fish migration routes to assist target species migration, and yet, fish migration science and practice remain imperfect. This study aimed to predict the migration paths of Gymnocypris przewalskii at the river confluence in the Qinghai Lake basin. The authors proposed an approach for fish migration route prediction, which involves a hydrodynamic module, a habitat model and a fish migration module. The TELEMAC‐MASCARET system was used to simulate hydrodynamic conditions and statistical analyses were carried out. During the flood season, G. przewalskii migrates downstream, whereas during the migratory season, they migrate upstream. The results indicate that flow velocity was the most significant hydrodynamic parameter affecting fish migration behaviour. The optimal flow velocity range for the target fish species was between 0.7 and 1.7 m/s. The impact of water depth was only observed in low discharge situations. Besides, the temperature plays a vital role in determining fish migration and abundance. However, the results reveal that in the morning hours, the temperature exhibits a range of 10°C to 16°C, and in the noon, the average temperature ranges from 16°C to 19°C, with a maximum temperature reaching 23°C. These temperature variations enable fish to migrate towards the tributary offering a more favourable water temperature during route selection at a river confluence. The study concludes that future research should consider incorporating the swimming capacity of the focal fish species to provide insights into fish migration patterns.