There are several algorithms for high-voltage transmission line fault location, such as steady state signal based one/two terminal algorithm, travelling wave based one/two-terminal algorithm. Due to many uncertainties and interference in transmission system, any single fault location method has its inherent limitations. By introducing multi-sensor data fusion technique, a self-adaptive weighted data fusion fault location algorithm is proposed, whose feature is fully using all the location information from multi-source. This new algorithm uses a priori knowledge or simulation data to obtain the weight coefficients of each single location algorithm corresponding to different fault conditions. The weight coefficients are then used to do calculation of weighted data fusion for all single location result. And finally the reliable and accurate location result will be obtained. This algorithm is tested under various fault conditions on a 500 kV transmission system model by using PSCAD/EMTDC. Simulation results show that the proposed algorithm is less influenced by fault resistance, fault position, fault type, system operation mode and distributed capacitance, and naturally possesses higher reliability, accuracy and robustness.