Landscape features have been shown to strongly influence dispersal and, consequently, the genetic population structure of organisms. Studies quantifying the effect of landscape features on gene flow of large mammals with high dispersal capabilities are rare and have mainly been focused at large geographical scales. In this study, we assessed the influence of several natural and human-made landscape features on red deer gene flow in the Scottish Highlands by analysing 695 individuals for 21 microsatellite markers. Despite the relatively small scale of the study area (115 × 87 km), significant population structure was found using F-statistics (FST = 0.019) and the program structure, with major differentiation found between populations sampled on either side of the main geographical barrier (the Great Glen). To assess the effect of landscape features on red deer population structure, the ArcMap GIS was used to create cost-distance matrices for moving between populations, using a range of cost values for each of the landscape features under consideration. Landscape features were shown to significantly affect red deer gene flow as they explained a greater proportion of the genetic variation than the geographical distance between populations. Sea lochs were found to be the most important red deer gene flow barriers in our study area, followed by mountain slopes, roads and forests. Inland lochs and rivers were identified as landscape features that might facilitate gene flow of red deer. Additionally, we explored the effect of choosing arbitrary cell cost values to construct least cost-distance matrices and described a method for improving the selection of cell cost values for a particular landscape feature.