Previous studies have suggested submarine landslides as sources of the tsunami that damaged coastal areas of Palu Bay after the 2018 Sulawesi earthquake. Indeed, tsunami run-up heights as high as 10 m determined by field surveys cannot be explained by the earthquake source alone although the earthquake is definitely the primary cause of the tsunami. The quantitatively re-examined results using the earthquake fault models reported so far showed that none of them could fully explain the observed tsunami data: tsunami waveforms inferred from video footage and the field survey run-up tsunami height distribution. Here, we present probable tsunami source models including submarine landslides that are consistent with the observed tsunami data. We simulated tsunamis generated by submarine landslides using a simplified depth-averaged two-dimensional model. The estimated submarine landslide model consisted of two sources in the northern and southern parts of the bay, and it explained the observed tsunami data well. Their volumes were 0.02 and 0.07 km3. The radius of the major axis and the maximum thickness of the initial paraboloid masses and the maximum horizontal velocity of the masses were 0.8 km, 40 m and 21 m/s in the northern bay, and 2.0 km, 15 m and 19 m/s in the southern bay, respectively. The landslide source in the northern bay needed to start to move about 70 s after the earthquake to match the calculated and observed arrival times.[Figure not available: see fulltext.]. © 2020, The Author(s). Show More