Abstract: Day-time (08.30-05.30 h) time budgets were generated from 55 light and 55 draft late pregnancy mares housed in tie stalls from ten pregnant mares' urine (PMU) farms using continuous video recording. Equal numbers of light and draft mares were filmed on each farm during the months of February and early March. The actions recorded included eating, drinking, resting (standing and recumbent), standing active, and interactions between horses (aggressive and non-aggressive). In addition, the presence and duration of stereotypic behaviours such as cribbing, head bobbing, weaving, and wood/bar chewing were recorded. Light mares spent significantly more time feeding and significantly less time standing active and standing resting (P<0.05, Rank Sum Two Sample Test). However, the time budget of both groups fell within the range of previously published activity budgets of feral horses. Therefore, the differences noted may not be clinically relevant. Three light and two draft mares performed repetitive behaviours at a level that is considered stereotypic (at >5% of their daily time budget). There was no significant difference in the number of horses performing stereotypies between light and draft mares. When the time budgets of both light and draft mares who performed stereotypies were pooled, the activities did not differ significantly from their counterparts who did not perform stereotypies. Because of the overall low prevalence of stereotypies and the fact that time budgets were similar to free-range horses, we believe that the management practice of keeping large numbers of pregnant mares in tie stalls is rational and that the welfare of mares is sound. Furthermore, we did not see a behavioural justification for a bias in the weight class of horses used within this management system.