The identification of the threat radar signal is an important task of the radar reconnaissance system, and is one of the factors that affect the success or failure of the countermeasure. Considering the complex electromagnetic environment with the increasing number of radars, the diversity of signal modulation modes and the increasing signal density, we proposed a new method for threat radar signal recognition based on TOA folding clustering. The method firstly performs TOA folding and sorting of the intercepted pulse stream, then extracts the data with DBSCAN clustering algorithm, and finally performs the template matching recognition of the threat radar signal. The simulation results show that the method can sort out the target radar signal in complex environment, and can adapt to the complex modulation of radar signal.