OBJECTIVE To test the feasibility of mounting a randomised controlled trial comparing mifepristone-misoprostol versus dilation and evacuation (D&E) for midtrimester abortion. METHODS Pilot randomised controlled trial. METHODS University of North Carolina Hospitals, Chapel Hill, North Carolina. METHODS Women aged 18 years or older and without prior uterine operations who requested abortion at 14-19 menstrual weeks of gestation from January 2002 to January 2003. METHODS Participants received either mifepristone 200 mg by mouth followed in two days by vaginal then oral misoprostol (Aberdeen regimen) or D&E after one or two days of laminaria preparation. Care was provided by residents under faculty supervision. METHODS Enrolment rate and acceptability of and adverse events associated with methods. RESULTS The trial was stopped at one year because of slow enrolment. Of 47 women eligible for the trial, 29 (62%) declined participation, primarily because of a preference for D&E abortion. Among the 18 participants enrolled, nine were randomised to treatment with mifepristone-misoprostol and 9 to D&E. Compared with D&E, mifepristone-misoprostol abortion caused more pain and adverse events, although none was serious. CONCLUSIONS Our findings concerning acceptability and adverse events should be considered hypothesis-generating; they may prove useful in planning a larger randomised controlled trial. Such a trial will be difficult to mount in the US. Hence, we suggest that it be done in a setting where labour-induction abortion is the norm.