The incidence and amount of migration of epidural catheters was investigated in a prospective randomised study of 153 women who required analgesia in labour. Inward or outward migration occurred in 36% of patients. Inward migration by 1-3 cm occurred in 21 (13.7%) patients and outward migration by 1 cm or more occurred in 34 (22.2%); three (2%) catheters migrated out through the skin. There were significant positive correlations between outward migration and weight, body mass index, and depth of the epidural space. There was no relationship between migration and height, age, intervertebral space used or duration of catheterisation. Problems with epidural block were no more likely in patients in whom migration of 1 cm or more occurred compared with those in whom migration was limited to a maximum of +/- 0.5 cm. However, the pattern of problems was different. All cases of failed epidural block occurred in patients whose epidural catheter migrated outward by 2.5 cm or more. Unilateral blockade was not more likely if migration of 1 cm or more occurred.