[Objective] The stability of the Gaojiawan landslide in Qinghai Province was evaluated to provide insights for disaster prevention and control.[Methods] Based on the analysis of the basic characteristics of the landslide, the formation conditions were studied initially. Afterwards, the stability status quo was evaluated by using the transfer coefficient method and the Swedish arc method, and the deformation prediction model was constructed using the long-term and short-term memory neural network to obtain the stability prediction evaluation. Subsequently, comprehensive research on the stability was performed by combining the stability status quo evaluation results.[Results] The Gaojiawan landslide had the basic conditions for the formation of landslides, and it had the characteristics of a multistage giant scale, and hence prevention and control should be comprehensive. Simultaneously, the present scenario of the landslide was stable, but a local secondary landslide might cause loosing stability, and that stability could be further weakened under adverse conditions. In addition, according to the deformation prediction research results, the average relative error was less than 2%. The results showed that the deformation of the Gaojiawan landslide would continue to increase, and there was no sign of convergence.[Conclusion] The stability of the Gaojiawan landslide could further weaken and tend to develop in a disadvantageous direction. Necessary measures should be taken as soon as possible to ensure stability.