Abstract: Mirror Self-Recognition (MSR) unveils complex cognitive, social and emotional skills and it has been found only in humans and few other species, such as great apes, dolphins, elephants and magpies. In this pilot study, we tested if horses show the capacity of MSR. Four subjects living socially under naturalistic conditions were selected for the experiment. We adopted the classical mark test, which consists in placing a coloured mark on an out-of-view body part, visible only through mirror inspection. If the animal considers the image as its own, it will use its reflection to detect the mark and will try to explore it. We enhanced the classical paradigm by introducing a double-check control. Only in the presence of the reflecting surface, animals performed tactile and olfactory exploration of the mirror and looked behind it. These behaviors suggest that subjects were trying to associate multiple sensory cues (visual, tactile and olfactory) to the image in the mirror. The lack of correspondence between the collected stimuli in front of the mirror and the response to the colored mark lead us to affirm that horses are able to perceive that the reflected image is incongruent when compared with the memorized information of a real horse. However, without replication of data, the self-directed behavior towards the colored marks showed by our horses cannot be sufficient per se to affirm that horses are capable of self-recognition.