While numerous works have shown that the road network impacts forest biodiversity, there is a lack of hindsight regarding how forest accessibility affects forest structure due to the possibility of human intervention through harvesting. In a context of growing demand for woody resources, it is important to quantify the link between accessibility and forest structure in order to anticipate a potential imbalance between accessible and remote areas regarding silviculture intensity. With the development of airborne laser scanning (ALS) assisted inventories, it is now possible to map the spatial variability of forest stand parameters at a resolution of about 20 m. Besides, Geographical Information Software (GIS) based models are now available to quantify forest accessibility based on the forest network and digital terrain models. The objective of this study is to quantify the link between forest parameters and accessibility, measured as the harvesting distance, while taking into account external environmental and human factors. The study area is located in the Vercors mountain (French Alps), where ALS data were acquired over 50 km2. Dominant species are Norway spruce (Picea abies), silver fir (Abies alba) and European beech (Fagus sylvatica). The area is mainly harvested by skidder, and presents a dense network of skid trails and forests roads. ALS prediction models were calibrated thanks to 96 field plots and used to map basal area, dominant height and quadratic diameter at 25 m resolution. Slope and altitude were computed from the ALS data. Meteorological and geological data were also gathered. Forest accessibility was mapped thanks to a GIS-based model (Sylvaccess) which outputs for each forest pixel the distance traveled by skidders to the nearest forest road. Preliminary data analysis showed that outside accessibility, the main factors affecting forest variables in this area are topographical factors, geology and forest ownership. Analyses of covariance (ANCOVA) were then performed with forest parameters as the dependent variable and slope, altitude, distance as the covariates and geology as categorical independent variable. The analysis was conducted separately for private and public forests as the distribution of forest parameters are significantly different for these two categories. For public forests, the adjusted R2 is 0.42 for the basal area ANCOVA model. All covariates have significant partial p-values (p