To improve the efficiency of safety tests of driver-automation cooperation, a method for generating a scenario library is proposed that considers the probability of scenario occurrence and driver-handling challenges in real driving situations. First, the original scenario data under cut-in conditions stored in a time series are extracted from the scenario data set. Then, a mathematical performance index is used to model the scenario and a significance function in terms of the occurrence frequency of the scenario, and the performance challenge between the driver and the vehicle is established. Next, the important scenario set is extracted from the original scenario set by constructing and optimizing a significance auxiliary function. Finally, the extracted important scenario sets are filtered by using the significance function values of the scenarios to generate a scenario library. Simulation results show that the proposed method for scenario library generation can effectively identify scenarios with potential adventure during driver-automation cooperation and thus accelerate safety tests compared with traditional methods.