•Regression modeling of soil loss on the AT identified seven significant factors.•Incorporation of GIS and LiDAR-derived variables increased explained variation.•Precipitation, landform slope, trail grade, and soil substrates were most influential.•Soil loss can be minimized by designing side-hill trails with lower grades. Recreational activities can negatively affect protected area landscapes and resources and soil erosion is frequently cited as the most significant long-term impact to recreational trails. This study applied extensive multiple regression modeling of trail soil loss to identify influential factors that managers can manipulate to improve the sustainability of trail design and management. Field measurements assessed soil loss as the mean vertical depth along 135 trail transects across the Appalachian Trail sampled along three 5 km trail segments in New Hampshire’s White Mountains National Forest, chosen due to its exceptionally high use and impact. GIS and LiDAR data were used to create many new variables reflecting terrain characteristics that were expected to influence trail erosion and improve predictive models of trail system soil loss. A variety of terrain and hydrology characteristics were applied to model trail soil loss at three spatial scales: transect, trail corridor, and watershed. The model for each spatial scale and a combined model are presented. The adjusted R2 explaining variation in soil loss is 0.57 using variables from all spatial scales, improving on predictive modeling from earlier studies. Environmental and trail design factors such as slope and watershed flow length were found to be significantly correlated to soil loss and have implications for improved sustainable trail design and management.