Abstract: Source location accuracy would inevitably deteriorate in the presence of observer position errors without taking them into account in multi-plane passive location system. An algorithm for multi-plane passive location that takes the observer position errors into account is proposed in order to solve the location problem in the presence of observer position errors. The proposed algorithm introduces the error correlation matrix of the argumented coefficient matrix in the observation equations caused by the measurement errors and observer position errors into the constraint, and uses the constrained least squares minimization on the errors of pseudo linear observation equations, which turns out to be equal to the generalized eigen-decomposition to a pair of matrix pencil. Simulation result indicates that the proposed algorithm achieves higher speed of convergence and high location precision, weakly affected by the observer position errors, and keeps good performance even in the condition of large measurement errors when compared with LS algorithm and EKF algorithm.