Despite a long history of human-horse relationship, horse-related incidents and accidents do occur amongst professional and non professional horse handlers. Recent studies show that their occurrence depend more on the frequency and amount of interactions with horses than on the level of competency, suggesting a strong need for specific research and training of individuals working with horses. In the present study, we review the current scientific knowledge on human-horse relationships. We distinguish here short occasional interactions with familiar or unfamiliar horses (e.g. veterinary inspection) and long-term bonds (e.g. horse-owner). An important aspect of the horse-human relationship is to try and improve the development and maintenance of a strong positive relationship. Studies show that deficits in the management conditions (housing, feeding, possibilities for social contact, and training methods) may lead to relational problems between horses and humans. Different methods have been used to assess and improve the human-horse relation, especially at the young age. They reveal that the time and type of contact all play a role, while recent studies suggest that the use of familiarized social models might be a great help through social facilitation. We argue that an important theoretical framework could be Hinde's [Hinde, R., 1979. Towards Understanding Relationships. Academic Press, Londres] definition of a relationship as an emerging bond from a series of interactions: partners have expectations on the next interaction on the basis of the previous ones. Understanding that a relationship is built up on the basis of a succession of interactions is an important step as it suggests that attention is being paid to the "positive" or "negative" valence of each interaction as a step for the next one. A better knowledge of learning rules is certainly necessary in this context not only to train the horse but also to counterbalance the unavoidable negative inputs that exist in routine procedures and reduce their impact on the relationship. It appears clearly that research is needed in order to assess how to better and safely approach the horse (e.g. research in position, posture, gaze, etc.), what type of approaches and timing may help in developing a positive bond, what influence human management and care have on the relationship, and how this can be adapted to have a positive influence on the relationship. Also the interaction between rider and horse, the search for the optimal match between two individuals, is an aspect of the horse-human relationship that requires attention in order to decrease the number of horse-riding accidents and reduced states of welfare. On the other hand, adequate knowledge is readily available that may improve the present situation rapidly. Developing awareness and attention to behavioural cues given by horses would certainly help decreasing accidents among professionals when interacting. Scientists therefore should play a major role in transmitting not only elements of the current knowledge of the ethology of the horse but also by helping developing observational skills. (c) 2007 Elsevier B.V. All rights reserved.