To integrate multiple location estimates obtained at different time instants by a three-satellite TDOA-based passive location system, a method based on sub-covariance weighting is proposed. Based on the outcomes of the location precision analysis of a single estimate, the algorithm yields an integrated estimate of the spatial location through a two-step process: firstly, by using the inverse of the sub-covariance as the weighting matrix, it estimated two of the three coordinates through weighted least square (WLS); secondly, it calculated the last one coordinate by solving the WGS-84 ellipsoid model function of the earth. Simulations show that the proposed method, compared with the existing methods, can significantly increase the location precision.