Abstract: Most of natural sands contain significant amounts of fines. However, the liquefaction discrimination methods for fines-containing sandy soils are always as a subsidiary to those for clean sands, and enough attention hasn't been paid to. The Chinese method and NCEER method are reviewed, and the discrepancy that two methods define clean sand for Chi-Chi earthquake data is found. By comparing the identification results of soils in Chi-Chi earthquake, the following improvements are suggested for the two methods: (1) for the Chinese method, firstly we should abandon the view that the clay content for silty sands is always 3%. Secondly, for the silty sands and silts with clay content not greater than 3% and fine content more than 15%, take ρ_c=F_c/4, if not, take 3%; (2) for the NCEER method, if the clay content is not greater than 3%, we should adjust (N₁)₆₀ to (N₁)_60cs only for the soils with fine content more than 20%. In doing so, the Chinese method is not any more too conservative, and the NCEER method are no longer tending to be dangerous for the soils with clay content not greater than 3% and fine content more than 5%.