In order to solve the problems of the least-square method in satellite fault detection,a Weighted Least-Square (WLS) algorithm is proposed for satellite fault detection and recognition.The algorithm uses the diagonal-line factor in the covariance matrix of the measurement equation as the weighting factor.Then the satellite-fault-detection algorithm based on the WLS method is obtained.The test statistic is established by using the quadratic sum of the pseudo-range residual errors of each satellite.The detection threshold is obtained from the false alarm rate of the fault detection and the probability density function.The calculated quadratic sum of the pseudo-range residual error is compared with the detection threshold,and then the fault detection and isolation is realized.Analysis is made to the performance of the two algorithms according to the measured data.The results show that the WLS RAIM algorithm is superior to the least-square RAIM algorithm on the fault detection sensitivity and fault recognition performance.