Despite the bacteriostatic effect of local anaesthetics, an epidural catheter causes bacterial contamination/infection, and we cannot therefore rely solely on this bacteriostatic effect. During continuous epidural analgesia, most bacteria appear to enter from a skin surface into the epidural space along the outside of an epidural catheter, since Staphylococcus epidermidis and S. aureus, which constitute epidermal bacterial flora, are frequently found in positive cultures of epidural catheter tips [1]. Therefore, bacterial culture of the outside, but not inside, of the catheter tip is important. In a tunnelled catheter with a subcutaneously implanted reservoir, however, the catheter system is isolated from a skin surface [2]. Furthermore, they described intraspinal infection, whereas we mentioned one case of catheter-tip contamination without infection. Therefore, these two results cannot be compared. Cefazolin is administered clinically at a dose of 1–4 g in an intravenous infusion, and the blood or tissue concentration may reach a similar or higher level to the epidural solution, 18 µg.ml−1. Therefore, the epidural cefazolin should not be neurotoxic. Of course, we do not think that epidural infection is entirely prevented by the cefazolin, and the skin was sterilised every other day. Usually, broad-spectrum antibiotics, including cephazolin, are beneficial for prophylaxis of contamination/infection, and can be administered by both the epidural and the intravenous routes. An epidural antibiotic is low cost as compared with an intravenous antibiotic.