Earthquake-Induced Landslide (EQIL) inventories are the key to improve our understanding of the relationship between landslides and their causes, including environmental settings and ground shaking parameters. However, creating a high-quality inventory can take years. As a result, reliable information on landslide-affected areas typically remains unknown until a complete inventory is compiled. In this paper, we analyze 20 digital EQIL inventories of varying quality and completeness that represent a range of geologic and climatic settings around the globe. We examine the landslide-affected area with respect to Peak Ground Acceleration (PGA) and topography, and develop a statistical model to estimate the landslide distribution without prior knowledge of the actual landslide triggering locations. For each EQIL inventory, we initially calculated the PGA contours where 90% of the total landslide population fell into. Subsequently, we define landslide susceptible areas as those pixels with slope > 5° and local relief>100 m. The latter is used to normalize the total landslide-affected area and to compute correlations with local PGA values. We find that the landslide-affected area may be predicted from PGA values only, with the mean error ranging from −20.0% to +7.1%, with respect to total landslide population. This relationship can be used immediately following a disaster to identify areas of greatest landslide impact and to prioritize emergency response actions, even without a landslide inventory.