Leader-follower behaviours have been described in cattle housed on pasture and when entering the milking parlour. To our knowledge, no study has investigated these behaviours in cattle provided free-choice access to the outdoors. This study explores whether the timing of cows exiting and entering the barn from an outdoor area is associated with the movement of their herd mates. We followed 18 lactating Holstein cows housed indoors in a freestall barn together with 18 non-experimental cows, for on average 77 d (range: 74 – 80). The 18 experimental cows were allowed free access to an outdoor open pack (sand covered with woodchips) through an automated selection gate which automatically recorded when a cow exited the barn. To enter the barn from outside, cows went through a one-way gate; time of entry was recorded from video. Cows moved in and out of the barn on average 269 times (range: 68 – 472). A Gaussian mixture model was applied to the log of the frequency distribution of time intervals between each cow and the next for each entry and exit. On the basis of where the curves intersected, we identified a threshold interval of 42 s between consecutive exits or entries to consider movements as associated. Applying this threshold, 13.8% of events were associated with the movement of another cow: 11.2% of exits and 16.4% of entries. We created a leader-follower matrix for each direction: the leader being the cow initiating the movement, and the follower the one walking through the same gate within 42 s of the leader. These association matrices were used to build weighted and directed social networks for exits and entries; we found a moderate positive correlation between the networks (RQAP = 0.69, p < 0.001), meaning the patterns of association between cows observed when exiting the barn were consistent with those when they entered. Just over a third of the relations in the networks were asymmetrical; in contrast, leader-follower theory would predict that one individual within each pair would lead consistently. We identified 17 of 153 possible between-cow pairings as preferentially bonded; these pairs had a simple ratio index of association more than twice the average of all possible pairs. We conclude that some of the cow movements to and from an outdoor pack were influenced by the movements of other group members, and these associations were particularly strong between some pairs of cows.