The increasing use of ambulatory surgery requires methods of anaesthesia that allow patients to be discharged soon after the operation is completed. Spinal anaesthesia is usually simple and quick, and the incidence of post-spinal headache has been reduced by using non-cutting small-gauge needles. Limiting the spread of spinal anaesthesia, as long as it still provides analgesia for surgery, should reduce the haemodynamic effects and speed recovery. Restricted spinal anaesthesia, intended to be unilateral using 0.18% hypobaric bupivacaine via a 25G or 27G Whitacre unidirectional needle, was compared with epidural anaesthesia (using a mixture of lignocaine and prilocaine) in 64 matched-pair patients undergoing ambulatory arthroscopy. Motor blockade, assessed for the specific myotomes L2 to S1, was significantly more unilateral in the spinal group. Two patients in the spinal group and nine patients in the epidural group were treated for hypotension (P < 0.05). One patient in the spinal group developed a post-spinal headache. One patient in the epidural group rated the anaesthesia poor.