Abstract: Between-previous termgroup transfernext term of adult female previous termponies,next term Equus caballus, was investigated for three consecutive summers on previous Assateaguenext Island, Maryland, U.S.A. Both the previous terminternext term-band movements of individual previous termmares and the marenext term turnover rates of one-male “harem” bands were examined. Long-term previous termtransfersnext term occurred at rates ranging from 0·06 to 0·18 per previous termmarenext term per month. previous termMaresnext term with foals transferred more frequently than previous termmaresnext term without foals, but neither female age, pregnancy, nearest-neighbour distances nor dominance rank affected the likelihood of transferring. Band turnover rates were uncorrelated with the average frequency of previous termmare-marenext term aggression within the band, but new previous termmaresnext term entering a band suffered a transient rise in aggression received. Thus, female aggression did not encourage, and may have discouraged, previous terminternext term-band previous termtransfers.next term Older stallions and stallions who had held bands for 2 years or more had significantly larger and more stable bands. Fewer previous termmarenext term turnovers were seen in bands whose stallions tended to face their previous termmares,next term showed a relatively high proportion of time feeding, and showed a relatively low proportion of time involved in aggression with other stallions, although at marginal levels of significance for all three variables. Thus, variability in stallion attributes, and possibly behaviour, probably plays the strongest role in determining previous termmare transfernext term patterns at previous termAssateague.next term