Shallow landslides are significant natural hazards in Oregon, and identification of areas susceptible to future landslides is a critical step in reducing risk. Recent advances in identification of areas susceptible to shallow landslides are mostly based on geographic information system (GIS) calculations of the slope stability using the infinite slope equation. This technique was further improved with high-resolution light detection and ranging (LiDAR)–based digital elevation models (DEM) converted to very accurate slope data as input into the GIS models. However, these models still underestimate and overestimate the susceptibility in certain areas compared to past landslide events and field observations. One significant overestimation we noted occurs in regionally flat areas with isolated steep slopes that have very little relief. We developed a method to remove these isolated overestimated areas using a neighborhood analysis with a maximum relief of 1.22 m (4 ft). Because landslides that originate on the steep slope may extend back into the flat area above the slope, or out onto the flat area at the toe of the slope, we applied a 9 m (30 ft) buffer (twice our defined depth to failure for shallow landslides) for all of the areas with a calculated factor of safety (FOS) less than 1.5. We tested the methods on three landslide inventory databases examining two main criteria: (1) capture rate (overall and individual landslides) and (2) reduction in total map area susceptibility coverage while maintaining a high capture rate. We found the two methods maintained a capture rate between 90% and 99% while at the same time reducing the total map area susceptibility zones from 64% to 42%.