In recent years, the illegal signal source localization applications of the indoor environment have become rather significant to the public information security, radio management and anti-terrorism scenarios. However, the popular indoor positioning methods, such as fingerprint-based positioning technologies, are not able to deal with the illegal signals for lacking of the priori information. In this paper, a kind of Direction of Arrival(DOA) estimation based method for indoor illegal signal source localization is proposed. The received array data are processed by the Eigen Space DOA(ES-DOA) algorithm to obtain azimuth spectrum and signal power estimation. A novel direct-path recognition approach is proposed to identify the real bearing from the azimuth spectrum. The position of the illegal signal source could be estimated by multiple crossover measurements. The simulation result shows that the ES-DOA algorithm could reach high resolution and has strong anti-noise capability dealing with the coherent array signals, and the proposed direct-path recognition algorithm shows high reliability and robustness in different indoor environments. The realization of the proposed indoor illegal signal source positioning method is convenient and flexible that the deployment could be in single station mode or multiple stations mode.