Abstract: Fishways can be one of the most important engineering measures to assist the fish in migrating upstream over a dam. However, the initial construction of fishways cannot consider the combination of hydraulic characteristics and fish behavior, leading to low effectiveness. The structure of the fishway can be optimized to evaluate the relationship between fish behavior and hydraulic factors. It is very necessary to develop an effective mathematical model, and then simulate the fish upstream migratory. In this study, a prediction model of upstream fish behavior was proposed to consider the multiple hydraulic factors and variable-speed swimming of Schizothorax wangchiachii(S. wangchiachii). The behavior data of S. wangchiachii was obtained from physical experiments and the euler-lagrangian-agent method (ELAM). The flow velocity, turbulent kinetic energy, vorticity, and strain rate were taken as the key hydraulic driving factors on fish movement. Firstly, principal component analysis was made to calculate the effects of flow velocity, turbulent kinetic energy, vorticity, and strain rate on the upstream behavior of S. wangchiachii. Next, the preference range of S. wangchiachii of these factors was determined through statistical analysis. Then, a fish behavior model was developed to combine the fish movement, including upstream behavior (the direction of fish movement was opposite to the direction of water flow), obstacle avoidance behavior (the fish movement was used to avoid some obstacles during movement), inertia behavior (the fish moved forward some distance due to the inertia when passing through the vertical seam), and random behavior (the random movement of fish under natural conditions). Additionally, a variable speed model of fish swimming was established to determine the response relationship between water flow velocity and fish swimming speed. Finally, the improved model was used to simulate the upstream trajectory of fish in the vertical slot fishway. Both simulated and measured trajectories were calculated using ninth-order polynomial regression. The reliability of the model was then verified to compare the characteristic trajectories. Four schemes of control experiments were designed to clarify the impact of vorticity, strain rate, flow velocity, and turbulent kinetic energy on fish upstream behavior. The preference range of hydraulic factors was then determined to tailor the influencing weight of vorticity, strain rate, turbulent kinetic energy, and flow velocity in the model. The experimental results indicate that the S. wangchiachii shared distinct preference ranges in the flow velocity, turbulent kinetic energy, vorticity, and strain rate. There was a relatively small error between the simulated and measured trajectory (R² = 0.935, RMSE = 0.043 m). As such, the upstream behavior of S. wangchiachii was accurately predicted in the fishway. There were different effects of vorticity, strain rate, turbulent kinetic energy, and flow velocity on the upstream behavior of S. wangchiachii. Furthermore, the R² of the simulated and measured trajectory were 0.526, 0.799, 0.861, and 0.910, respectively, when considering the vorticity, strain rate, turbulent kinetic energy, and flow velocity on the fish upstream behavior. Correspondingly, the RMSE values were 0.127, 0.09, 0.074, and 0.054 m, respectively. All errors were also lower than those observed in the original. Therefore, there was a significant influence of the above four hydraulic factors on the fish tracing. The weight coefficient of vorticity was 0.246, which was relatively small. But it played a key role in the process of fish upstream, in order to avoid the missing fish. The prediction model of fish upstream behavior can provide a strong reference to design and optimize the fishways.