Radiological evidence of the distribution of solutions in the epidural space indicates that the volume used and the site of injection are the most relevant factors, while the height of the patient is of little importance, and the rate of injection, posture, and age of the patient exert no influence. Even so, there are wide variations in spread with a given volume via the same route, and it is impossible to predict accurately the level which will be attained. Clinical improvement following epidural injection for the lumbosciatic syndrome does not appear to be correlated with a wide dissemination of the solution used, and large volumes seem to confer no advantage. The caudal route perhaps deserves re-evaluation, though the small number in this series prevents valid conclusions being drawn.