NTS A HIGH IMPEDANCE Fault (HIF) detection METHOD BASED ON FUZZY CLUSTERING. THE MAIN BENEFITS OF THIS METHOD ARE ITS SIMPLE STRUCTURE, FAST LEARNING AND ITS INDEPENDENCE OF EXPERT KNOWLEDGE. MOREOVER, IT CAN BE TRAINED BY SOME SAMPLING DATA. THE FEATUERS OF PHASE CURRENT, WHICH ARE THE INPUTS OF FUZZY SYSTEM, ARE CURRENT MAGNITUDE OF FIRST, THIRD, FIFTH, DEFERENCE BETWEEN THIRD AND FIFTH HARMONICS, PHASE OF THIRD HARMONIC AND MAGNITUDE OF ZERO SEQUENCE OF FIRST HARMONIC. TRAINING DATA SET IS OBTAINED BY SIMULATION OF ELECTROMAGNETIC TRANSIENTS PROGRAM(EMTP) ON NORMAL CASES SUCH AS SWITCHING CAPASITORS, NORMAL LOADS AND Fault CASES LIKE FaultS ON SANDY SOIL, ASPLALT, DRY SOIL. RESULTS OF HIF detection FOR SOME CASES ARE PRESENTED, AND SHOW THAT THE HIF detection CAN BE PERFORMED IN A VERY EFFICIENT MANNER BY USING THIS ALGORITHM.