The prescriptive re-compaction method has been used as a common technique for upgrading loose fill slopes in Hong Kong. Due to the heavy plants required for the re-compaction work, the method may be extremely difficult and dangerous to be applied in many congested areas. To explore the possibility of using soil nails to tackle with the potential problem of liquefaction of loose fill slopes, experimental and numerical investigations have been carried out. The experimental work aims at studying the liquefaction behaviour of loosely compacted decomposed granites, which have been commonly used as a filling material. Undrained triaxial tests have been conducted on two decomposed granites with similar fines content. The use of fines content as a quantifier of the liquefaction potential, which is defined as the percentage drop of deviator stress during undrained strain softening, has been examined. The influence of moisture content at preparation on the liquefaction potential of decomposed granites has also been studied. In the numerical analyses, a 20 m high 35° loose fill slope was set up using a finite difference code. By employing a "pseudo" static liquefaction soil constitutive model, the behaviour of loose fill slopes with and without soil nails subjected to the effect of surface infiltration has been investigated under different assumed conditions. Reverse behaviour reported in silty sand is observed in decomposed granites which is characterised by an increase of liquefaction resistance as the confining pressure is increased. Moreover, liquefaction potential of a soil specimen is dependent on the moisture content at preparation. When a specimen is prepared at moisture contents drier than the optimum, a lower liquefaction potential would result as compared to specimens prepared at optimum moisture content. Fines content alone is found not to be a good measure for quantifying the liquefaction potential of soils. An empirical method to estimate the liquefaction potential is proposed. The method is based on the ...