Target threat assessment is a key problem in air combat situational awareness. Aiming at the defects of the traditional methods that are difficult to be both accurate and realtime, a method based on the extreme learning machine_ adaptive boosting (ELM_AdaBoost) strong predictor is proposed. Combined with the AdaBoost classification algorithm, the ELM algorithm is improved and the ELM_AdaBoost algorithm is proposed to establish the strong predictor. The air combat data is selected in the air combat maneuvering instrument and the sample data for target threat assessment is established using the threat index method. A target threat assessment model based on the ELM_AdaBoost strong predictor is established and the number of hidden layer nodes of ELM network and the weaker predictors is determined respectively in a certain range which makes the algorithm have better predictive accuracy. The accuracy analysis and realtime analysis of the assessment are carried out in simulation experiments and the results show that the proposed method reduces the time spent in the assessment while ensuring a higher degree of accuracy, which can achieve accurate and rapid target threat assessment in air combat.