To develop a method for objective assessment of equine post-operative pain. Prospective nonblinded clinical study. Twelve adult horses: Group 1 (G1, n = 6), admitted for arthroscopy (under general anaesthesia, with multimodal analgesia); Group 2 (G2, n = 6), 'pain free' controls. Horses were filmed continuously (CVI, time-lapse video recorder) over 72 hours, from 24 hours pre-surgery (PS) to 48 hours post-recovery (PR) (G1), and over 24 hours (G2). Activity budgets were determined from 24 to 0 hours PS, 0-24 and 24-48 hours PR (G1) and for 24 hours (G2). Using direct observation (DO), active behaviours and postures were recorded at set time points PS and PR (G1) and at two time points (morning/evening) (G2). Heart rate (HR) and respiration rate (RR) were recorded simultaneously. Statistical analysis investigated within-group and between-group time-related changes in behaviour, HR and RR. There was no difference in HR or RR between G1 and G2 at any time point. Anaesthetic 'hangover' and hunger-related activity modulated behaviour from 0 to 6 hours PR, when abnormal postures and locomotion occurred in all G1 horses, but no G2 horses. Compared with G1 (0-24 hours PR), G2 spent more time eating (t = -3.34, p < 0.01), more time at the front of the stable (t = -2.42, p < 0.05), and less time exhibiting 'abnormal' behaviour (U = 56, p < 0.01). Comparing PS and PR behaviour, G1 spent less time exploring from 24 to 48 hours PR (t = 3.49, p < 0.05), less time in sternal recumbency (t = -3.8, p < 0.05) and less time moving (t = 3.19, p < 0.05). Horses tended (p < 0.07) to spend less time positioned in the front of the stable PR (less from 24 to 48 hours PR than from 0 to 24 hours PR). Comparing PR (evening) behaviour, G2 spent more time with head above withers (U = 21.5, p < 0.01), and ears forwards (U = 22, p < 0.01). G1 showed time-related changes (all p < 0.05) in time with lower lip tense (S = 15.8), eating (S = 17.08) and with head positioned above withers (S = 18.04). No differences in event behaviours were observed between G1 and G2. Within G1, only olfactory behaviour varied significantly with time (S = 14.52, p < 0.05). Changes in equine behaviour suggestive of post-operative discomfort were identified using both DO and CVI. Analysis of activity budgets may be a more sensitive method of identifying behavioural changes indicative of equine discomfort than repeated DO of specific events and postures.