Veterinary care can be essential for maintaining the physical health of horses and yet the impact of veterinary care on a horse's emotional state and how that influences their behavioural responses has not been investigated previously. The term horses is used to describe both horses and ponies. Adverse behavioural responses are unwanted, not only for equine welfare but because veterinarians are frequently injured by the horses they are working with, each sustaining on average around 7-8 significant injuries over the courses of their careers. This study evaluated which aspects of veterinary care horses find stressful, their common behavioural responses in this context and potential ways of improving how they perceive veterinary care. In the context of a veterinary examination, horses are limited in which behavioural responses they can express as they will be restrained to a greater or lesser degree. To determine the range of behavioural responses that may occur and their usefulness in assessing emotional state in the veterinary clinic, a two round Delphi process was undertaken. In the first-round participants from the field of Equitation Science (n=16), Equine Veterinarians (n=10) and Animal Behaviour and Welfare Scientists (n=7) evaluated 11 videos of horse's behavioural responses to veterinary care. They assigned a score to the degree of stress they perceived each horse to be experiencing and described any behavioural indicators of stress as free text. There was poor agreement on the perceived degree of stress the horses were experiencing and different behavioural indicators were reported by different individuals. The behavioural indicators (41) identified in round one were then re-presented to the participants in a second round where 34/41 indicators were considered useful measures of the horse's stress response in this context. A feasibility study was then undertaken to determine the best measures of arousal and valence in the context of a veterinary clinic. Horses presenting for veterinary care at an equine hospital (n=30) were filmed during various veterinary interactions (Events) with the horses and the footage was evaluated using a detailed inventory of behavioural responses. Principle component analyses were performed, results of which suggested that the most common behavioural responses to veterinary care were characterised by freeze or flight. Evaluation of the feasibility of different measures of arousal (salivary cortisol, surface eye temperature and heart rate variability) were undertaken on the same cohort of horses. Salivary cortisol concentrations highlighted that horses were in a state of elevated arousal, likely as a consequence of transportation, even before veterinary care commenced. Elevated cortisol concentrations were maintained, most likely due to exposure to veterinary Events, but the lag between an individual Event and concentration peaks / return to baseline meant this was not feasible measure in this context. Surface eye temperature provided data that varied in real time and showed variation between horses and between individual Events. The main limitation of using this measure was the practical challenges trying to capture usable, in focus, and therefore accurate, thermal images of the eye as many horses did not remain still during Events. Heart rate variability parameters (HRV) provided data that varied in real time and showed variation between horses and between individual Events. Moreover, it could be collected remotely and regardless of any behavioural response. In the feasibility study HRV data were collected using a Polar heart rate monitor. However, validation against an ECG showed the Polar to be inaccurate, even following artefact correction. Even when an ECG was used subsequently, manual R peak detection and documentation of arrhythmias was essential to ensure accurate data for HRV analyses were generated. The methodology developed in the feasibility study was used to evaluate the responses of 60 horses undergoing veterinary care in the same equine hospital. The proportion of time a horse was 'stressed' was calculated based on the sum total of time in any negative affective state over total time. This study found marked variation regarding which veterinary Events horses appeared to find stressful, both within and between individuals. For example, 13/15 veterinary Events resulted in a range of time an individual horse was scored as 'stressed'. Negative behavioural responses were characterised by freeze or flight but also fidgeting behaviours, which are well described in other species but poorly documented in horses. Aggressive 'fight' responses occurred very infrequently. Further research is required to determine whether training of equine veterinarians to recognise subtle behavioural indicators, across the range of responses (freeze, flight, fidget or fight), can be used to improve equine welfare and reduce the likelihood of occupational injuries. With regard to evaluation of individual Events: • One of the most arousing stimuli, based on behavioural response and elevated heart rate, in the veterinary environment was the sound or sight of another horse walking past and so minimising this stimulus where possible is recommended. • Horses are frequently walked onto a weigh bridge upon entering the equine hospital and this commonly evoked a behavioural response characterised by freeze or flight and elevated heart rate. It is less than ideal that their first experience in a novel environment and first interaction with staff may be a negative one. Weighing a horse after other veterinary Events have been completed or development of a low stress technique would likely be of benefit. • The use of feed during diagnostic local anaesthesia appeared to minimise the proportion of time an equid was scored as 'Stressed' compared to when it was not used and so is recommended. • Horses were scored as more 'Stressed' when an I.V. injection was administered by a student or nurse in comparison to a veterinarian. Investigation of training to try to mitigate this response should be undertaken. The final study in this thesis evaluated the way in which learning theory could be applied in veterinary practice. Classical counter conditioning was used prior to diagnostic local analgesia (nerve blocks) and the impact on the horse's stress response was investigated. Twenty-seven cases presenting for lameness evaluation were recruited and randomly assigned to a control or treatment group. In the treatment group the provision of feed (unconditioned stimulus) was paired with someone approaching (conditioned stimulus) the limb. In the control group 14 horses received 24 nerve blocks, in the treatment group 13 horses received 35 nerve blocks. Restraint during the nerve block was at the discretion of the veterinarian performing the procedure. Video footage of horses during the nerve block was evaluated using a fixed list method Qualitative Behaviour Assessment by 5 observers blinded to treatment group. They scored the treatment group as being in a more positive affective state (at ease, relaxed and confident) than the control group (nervous, tense and fearful). However, the difference between groups was only statistically significant once other significant contributing factors such as the type of block were included in the multivariable model. This highlights that many different factors are contributing to how a horse perceives veterinary Events such as nerve blocks but that classical counter conditioning can be used to help horses perceive them more positively. In summary, a horse's perception of veterinary care will vary markedly between individual animals but also within an individual. Constant evaluation of the horse's emotional state, based on an understanding of subtle indicators of negative affect, will allow the veterinary team to modify their approach and optimise care of the individual. In addition, it is possible to promote a positive affective state even during aversive procedures such as nerve blocks. This can be achieved in a real-life scenario using classical counter conditioning.