Coastal marten (Martes caurina humboldtensis) require multiple structures (e.g., trees, snags, logs) within home ranges for resting and denning, often serving as locations to avoid predators or mitigate adverse weather conditions. Land managers urgently seek vegetation information to identify risks between fuel-focused forest treatments and conservation of essential areas used by martens. We identified 88 resting locations (n = 18 martens) using both fine-scale GPS/accelerometer data or VHF telemetry in southern Oregon. We collected plot-level conditions (basal area, shrub and horizontal cover, woody material, slash/rock pile volume, canopy cover, foliage height diversity, plant area volume density) at used and random locations (n = 76, 74, respectively). Preliminarily, area of both slash and rock piles were higher at used than random locations (slash: 165±1409 m3, 5±33 m3; rock: 9,422±81,317 m3,11±40 m3, respectively). Shrub and horizontal cover were also modestly higher at used than random locations (shrub: 52±20%, 39±23%; horizontal cover: 78±16%, 68±19%, respectively). However, our confidence intervals overlapped, and summarizing vegetation metrics resulted in weak correlations of marten use. Moving forward, we will use machine learning to build decision trees describing selection. Our results highlight challenges of describing coastal marten habitat even within a relatively homogenous study area.