Abstract: A novel fault type classification approach for ultra-high voltage (UHV) transmission lines was proposed based on the adaptive-network-based fuzzy inference system (ANFIS) to distinguish the ten common fault types, including single line to ground faults, line to line to ground faults, line to line faults, and three-phase fault. In this approach, the standard deviation and inter-quartile range of fault components of one cycle post-fault-current are taken as the characteristic quantities of fault classification. The influence of noise and harmonic on the characteristic quantities was analyzed. A fault classification model based on the ANFIS was established. A large number of simulations were carried out in PSCAD/EMTDC (power systems computer aided design/electromagnetic transients including direct current). The results indicate that the proposed approach is capable to identify fault types fast with a high accuracy up to 99.5%. Furthermore, the approach is insensitive to different fault initial angles, fault distances and fault resistances and has a good adaptability for different noise levels, harmonics, transform characteristics of current transformer (CT) and sampling frequencies.