Large mammalian herbivores have a disproportionately high impact on terrestrial ecosystems compared to other species and play unique ecological roles as environmental engineers. The (re)introduction of large herbivores is gaining momentum as an effective way to rewild and restore grassland habitats. Large herbivores, like horses (Equus ferus), are excellent candidates for rewilding activities, as they can contribute to reducing the loss of open landscapes. The rewilding based on horses relies on domestic individuals. Understanding the ecology and behaviors of such populations may become crucial to restore self-sustaining wild (feral) populations. This contribution presents preliminary results  of a study conducted on the population of feral horses inhabiting a mountainous range in Central Italy (La Calvana) recognized as SIC (IT5150001) of Natura2000 network. This feral horse population can be considered as a sort of unintentional rewilding process, as it originated from a small group of free-ranging horses abandoned in the ’70.  The aim of the study is to assess the structure of the population, to provide a description of its behaviors, and to quantify their potential impact on grassland habitats. Firstly, we analyzed population structure and collected samples for genetic analysis. Secondly, we sampled plant communities in two different areas of animal range, characterized by different levels of grazing intensity (90% vs 50%). Moreover, exclusion cages were installed in the two areas to quantify animal intake along the season. The results showed that the population is divided in 9 separate herds (each composed by a dominant male, several mares and a few foals or juveniles), plus some groups of bachelors composed by mature but non-reproducing males. Vegetation data highlights that horse grazing leads to a selection of plant communities in terms of biomass and species composition which could promote the conservation of grassland habitats threatened by shrub and tree encroachment.  These preliminary data will be used to model the population dynamics and provide preliminary estimates on its long-term effect on the environment.