Abstract. The Ms 8.0 Wenchuan earthquake in 2008 and Ms 7.0 Lushan earthquake in 2013 produced thousands of landslides in the southern region of the Longmen Mountains in China. We conducted field investigations and analyzed remote sensing data to determine the distribution law of earthquake-triggered landslides. The results show a strong negative power law relationship between the size and frequency of landslides in VII, VIII, and IX seismic intensity zones, a weak power law in the X seismic intensity zone, and a lognormal distribution in the XI seismic intensity zone. Landslide density increases with increasing seismic intensity. A sandpile cellular automata model was built under the conceptual framework of self-organized criticality theory to simulate earthquake-induced landslides. Data from the simulations demonstrate that, with increasing disturbance intensity, the dynamical mechanism of the sandpile model changes from a strong power law to a weak power law and then to a lognormal distribution. Results from shaking table experiments of a one-sided slope sandpile show that, for peak ground acceleration (PGA) in the range of 0.075–0.125 g, the relation between the amount and frequency of sand follows a negative power law. For PGA between 0.15 and 0.25 g, the relation obeys a lognormal distribution. This verifies that the abovementioned distribution of earthquake-induced landslides should be a universal law from a physical viewpoint and may apply to other areas. This new perspective may be used to guide the development of an inventory of earthquake-triggered landslides and provide a scientific basis for their prediction.