A healthy 22-year-old man received an initial injection of 12 mL of lignocaine/bupivacaine solutions (2 mL test, then 10 mL) into an epidural catheter. This produced a satisfactory regional blockade that seemed to be epidural but, when a supplementary 6 mL injection was given 1 h later, the patient developed impaired motor function as far as the upper cranial nerves, with loss of pinprick sensation to the shoulder. The emergence of fluid dribbling freely from the catheter prompted measurement of the pressure, which was 36 mmHg. The fluid was proved not to be cerebrospinal fluid (CSF) by the absence of glucose (on dextrostix), by the appearance of turbidity with added thiopentone, and later by microscopy. Slow aspiration of 7 mL of the presumed injectate reduced the pressure in the catheter to 8 mmHg, which promptly reversed the additional excessive blockade, allowing surgery to proceed uneventfully. The retrieval of injectate argues strongly that the catheter tip had found its way subdurally, and the promptness of the reversal with aspiration argues for a mechanical rather than a pharmacological cause for the extensive neurological dysfunction after the second injection. Pressure measurement and aspiration may be helpful in other similar cases.