Background. Caesarean birth rates in North America continue to rise, in the absence of benefit for mothers and babies. One reason may be that hospitalized labouring women spend most of their labours in recumbent or semi-recumbent positions. Although hands-and-knees position has theoretical advantages, efforts to encourage its adoption in practice are severely hampered by the lack of compelling evidence that it is beneficial. Before a definitive, large scale trial, with spontaneous vaginal birth as the primary outcome, could be justified in terms of time, effort, and expense, several feasibility and acceptability questions had to be addressed. We aimed to enrol 60 women in a pilot study to assess feasibility and acceptability of the trial protocol, and to obtain estimates of treatment effects on method of birth and persistent back pain. Methods. We conducted a pilot study at two North American hospitals. In ten months of recruitment, 30 nulliparous women in labour at term were randomly allocated to either usual care (use of any position during labour except hands-and-knees) or to try hands-and-knees for 15 min every hour during labour. Data were collected about compliance, acceptability, persistent back pain, intrapartum interventions, and women's views of their experiences. Results. Although mean length of time from randomization to delivery was over 12 hours, only 9 of the 16 women allocated to repeated hands-and-knees used it more than twice. Two of the 14 in the usual care group used hands-and-knees once. Twenty-seven women had regional analgesia (15 in the hands-and-knees group and 12 in the usual care group). Eleven in the hands-and-knees group and 14 in the usual care group had spontaneous vaginal births. One woman (in the hands-and-knees group) had a vacuum extraction. Four women in the hands-and-knees group and none in the usual care group gave birth by caesarean section. Hourly back pain ratings were highly variable in both groups, covering the full range of possible scores. Given the low compliance with the hands-and-knees position, it was not possible to explore relationships between use of the position and persistent back pain scores. When asked to rate their overall satisfaction with their birth experiences, the hands-and-knees group's ratings tended to be lower than those in the usual care group, although 11 in the hands-and-knees group and 8 in the usual care group stated they would probably or definitely try the position in a subsequent labour. Conclusion. We concluded that we could not justify the time and expense associated with a definitive trial. However such a trial could be feasible with modifications to eligibility criteria and careful selection of suitable settings.