Evaluation of ground liquefaction hazards (e.g. the triggering and the post-liquefaction consequences) is essential for the seismic design and operational safety of infrastructures. To investigate the evaluation capability of different ground motion intensity measures (IM), the present study conducts a series of dynamic centrifuge model tests under 50 g centrifugal acceleration. A level saturated sand model ground is shaken 20 times with different amplitudes of input motion to obtain both liquefied and non-liquefied case histories. Then the calculation procedures of the peak ground acceleration amax, the earthquake-induced cyclic stress ratio CSR, the Arias intensity Ia and the filtered accumulative absolute velocity CAV5 are presented based on the recorded base motion. All these IMs are processed and checked by the model test data. The analyses indicate that, several types of IMs have similar ability to evaluate the liquefaction triggering. There are obvious IM thresholds from the generation of excess pore pressure to the initial liquefaction trigger. Several types of IMs have differences in the evaluation ability of post-earthquake settlement. Ia and CAV5 perform better than amax and CSR, and the possible reasons are preliminary explained. This study presents basis for reliable evaluation of liquefaction catastrophe by selecting appropriate ground motion intensity index.