Cues such as the human pointing gesture, gaze or proximity to an object are widely used in behavioural studies to evaluate animals’ abilities to follow human-given cues. Many domestic mammals, such as horses, can follow human cues; however, factors influencing their responses are still unclear. We assessed the performance of 57 horses at a two-way choice task testing their ability to follow cues of either a familiar (N = 28) or an unfamiliar informant (N = 29). We investigated the effects of the length of the relationship between the horse and a familiar person (main caregiver), their social environment (living alone, in dyads, or in groups) and their physical environment (living in stalls/paddocks, alternating between paddocks and pastures, or living full time in pastures). We also controlled for the effects of horses’ age and sex. Our results showed that horses’ success rate at the task was not affected by the familiarity of the informant and did not improve with the relationship length with the familiar informant but did increase with the age of the horses. Horses living in groups had better success than the ones kept either in dyads or alone. Finally, horses housed in small paddocks had lower success than those living on pasture. These results indicate that with age, horses get better at following human-given indications regardless of who the human informant is and that an appropriate living and social environment could contribute to the development of socio-cognitive skills towards humans. Therefore, such aspects should be considered in studies evaluating animal behaviour.