The time domain based fault location method needs a short time window and can compensate for the deficiency of the frequency domain method. However， with the effect of uncertainty of transmission line parameters and the two terminals being asynchronous taken into account， time domain methods are inaccurate and unstable. Therefore， by utilizing the steady fundamental frequency phasors during the pre-fault normal state， a self-adapting observation equation is established to obtain the precise transmission line parameters and two terminals asynchronous time difference. Meanwhile， to solve this equation， the particle swarm optimization（PSO） and least squares iteration are combined. Then， a Bergeron modeling fault location time domain observation equation is set up and solved by the PSO algorithm. Consequently， a time-frequency-domain fault location scheme with two terminals being asynchronous and transmission line parameters self-adapting is presented. The simulation results concerning a 1 000 kV transmission system model show that the short-time-window method proposed is accurate， reliable and impervious to the uncertainty of transmission line parameters and two terminals being asynchronous.