Forty-eight Large White x Landrace multiparous sows were mixed into twelve groups of four animals after their piglets were weaned. These groups were defined as static, with no animals being added to or removed from the groups after their formation. Aggressive and submissive behaviours were recorded continuously for 9 h after the sows were mixed, and the sows were assigned high or low social status on the basis of their relative aggressiveness and success in aggressive interactions. After five weeks, each static group was mixed into a dynamic group of 40 ±2 sows for an 11-week period. Three static groups (ie 12 animals) at a time were added to the dynamic group at three-week intervals; the same number of animals was removed at these time-points in order to maintain the group number at 40 ± 2. Injury levels increased significantly with the transition from static groups to the dynamic group (P < 0.001). Sows with low social status had lower bodyweights (P < 0.001) and higher injury levels one week after mixing into static and dynamic groups (P < 0.01). Social status did not significantly affect salivary Cortisol levels. Sows with low social status were positioned lower in the feed order, determined using an electronic feeder (P < 0.001), and tended to be displaced from the feeder queue more often (P < 0.1) in the dynamic group. Sows with low social status were also displaced from the drinker more often than highranking sows in the dynamic group (P < 0.01). This may have led to the greater frequency of drinking behaviour shown by low-ranking sows (P < 0.05). Sows with low social status were observed less often in the kennel areas than were the high-ranking sows in the dynamic group (P < 0.05), suggesting that they were denied access to the prime lying areas. The results suggest that the welfare of sows is negatively affected by low social status in both small static and large dynamic groups.