One hundred women were randomly allocated to the left lateral, Oxford or sitting position for induction of combined spinal-epidural anaesthesia for Caesarean section using 2.5 ml hyperbaric bupivacaine 0.5% and 10 mu g fentanyl. Women in the left lateral were then turned to the right lateral position; women in the Oxford position were turned to the same position on their opposite side; and women in the sitting group were turned to the supine left tilt position. Women remained in these positions until ready for surgery, which was conducted in the supine position with a wedge placed under the right hip. Ephedrine requirements before re-positioning for surgery were less in the sitting position than in the other two positions: median (IQR [range]) doses for the lateral, Oxford and sitting groups were 21 (12-30 [6-48]), 18 (7.5-24 [6-48]) and 12 (6-21 [6-42]) mg, respectively; p = 0.04. Sensory block to touch sensation at the T5 dermatomal level was most quickly achieved in the lateral position with median (IQR [range]) block onset times for the lateral, Oxford and sitting groups of 9 (6-13 [4-30]), 15.5 (9-22 [4-34]) and 14 (9-18[6-36]) min, respectively; p = 0.004. In the Oxford position, more epidural catheters required dosing to achieve a sensory block of T5 before surgery: the number of patients (proportion) bolused in the lateral, Oxford and sitting groups was 1 (3%), 7 (22%) and 1 (3%), respectively; p = 0.01. We did not demonstrate any advantage in using the Oxford position for combined spinal-epidural anaesthesia for elective Caesarean section.