Roadkill is one of the most prominent causes of wildlife mortality. Much research effort has focussed on collisions with ungulates because of traffic safety. However, studies about large carnivore roadkills are scarce despite vehicles being a main cause of mortality. The absence of studies can be explained in part because of difficulties in obtaining sufficient sample sizes. We collected data from locations of 82 wolf roadkill sites in the Castilla y León Region, northwest Spain. We evaluated different models to characterise collision localities using logistic regressions with corrections for rare events. The best models included traffic and human disturbance parameters. Landscape variables did not improve predictive power. Fencing was a decisive key predictor; roadkill was proportionally higher along fenced highways than on similar major roads that lacked fences. Palabras clave: Wolf