||Mirror self-recognition (MSR), investigated in primates and recently in non-primate species, is considered a measure of self-awareness. Nowadays, the only reliable test for investigating MSR potential skills consists in the untrained response to a visual body mark detected using a reflective surface. Here, we report the first evidence of MSR at group level in horses, by facing the weaknesses of methodology present in a previous pilot study. Fourteen horses were used in a 4-phases mirror test (covered mirror, open mirror, invisible mark, visible colored mark). After engaging in a series of contingency behaviors (looking behind the mirror, peek-a-boo, head and tongue movements), our horses used the mirror surface to guide their movements towards their colored cheeks, thus showing that they can recognize themselves in a mirror. The analysis at the group level, which 'marks' a turning point in the analytical technique of MSR exploration in non-primate species, showed that horses spent a longer time in scratching their faces when marked with the visible mark compared to the non-visible mark. This finding indicates that horses did not see the non-visible mark and that they did not touch their own face guided by the tactile sensation, suggesting the presence of MSR in horses. Although a heated debate on the binary versus gradualist model in the MSR interpretation exists, recent empirical pieces of evidence, including ours, indicate that MSR is not an all-or-nothing phenomenon that appeared once in phylogeny and that a convergent evolution mechanism can be at the basis of its presence in phylogenetically distant taxa.