Objective, non-invasive indicators of the subjective experience of positive emotion are required to support assessment and improvement of animal welfare. Emotion is unique to the individual and indicators of emotion are indirect. The aim of this thesis was to ascertain if body and facial behaviours and physiological parameters reflected the emotional experiences of horses. Following review of the theoretical and experimental literature, three experiments were conducted, and an alternative emotional arousal-valence framework was proposed. Based on the preferences of individual horses, the relative arousal level and emotional valence induced by four stimuli (wither grooming, motionless person, intermittent spray, and being left alone) were ordered. Behavioural and physiological parameters were then measured during exposure to each stimuli. The indicators of contrasting affective experiences in horses were found to be heart rate, heart rate variability, eye temperature, and behaviours involving legs, neck, tail, ears, eyes, eyebrows, mouth, chin, and nares. Several behaviours differed across all three arousal levels or valence levels. Positive emotional valence was indicated by a decreased rate of neck very low, left ear forward, left or right ear back, blink, angled eyebrow, nares flared, nares neutral, and/or an increased rate of chin wobble, small eye aperture, or oral investigation behaviours. Higher arousal was indicated by an increased rate of neck very high, tail swishing, or higher odds of contracted lips, and/or a decreased rate of right ear forward or to the side behaviours. Reduction from 16 to six parameters may be possible. The findings may be used to aid interpretation of horse emotional experience and in the assessment and improvement of horse welfare. The research approach and framework described in this research may be suitable for future research in horses and other species.