Abstract: The aperture of the conventional double parallel uniform linear array cannot be effectively expanded due to the restriction of spatial sampling theorem. As a result， the accuracy and the Degree of Freedom （DOF） of its 2D Direction-of-Arrival （DOA） estimation cannot be effectively improved. Aiming that， a 2D-DOA estimation method base on the generalized coprime double parallel array is proposed. The method virtually expands the generalized coprime double parallel array to obtain two difference co-arrays， which contain have more virtual elements， and utilizes the correlation and cross correlation information of the co-arrays to construct an expanded matrix containing the 2D angular information. By employing an improved 2D Root-MUSIC algorithm to replace the conventional 2D peak search algorithm， the method automatically matches the incident signals and further obtains the corresponding DOA estimators estimates the DOAs. Simulation results demonstrate that the proposed method has lower computation complexity， better estimation accuracy and higher DOF than the conventional methods.