BACKGROUND Many cases of cauda equina syndrome after maldistribution of local anesthetics during continuous spinal anesthesia have been reported. In experiments, a caudad route of catheter travel and the use of hyperbaric agents have been shown to induce these limited blocks. The aim of this clinical study was to verify this hypothesis and seek a predictive factor for the maldistribution of bupivacaine. METHODS Continuous spinal anesthesia via a 19-gauge end port spinal catheter was performed in 80 elderly patients randomly assigned to receive either isobaric or hyperbaric solutions. Successive injections of 2.5 mg bupivacaine were performed at 5-min intervals until a sensory level at or cranial to T8 was obtained. Maldistribution was defined by a sensory level caudal to T12 despite a total dose of 17.5 mg of either isobaric or hyperbaric bupivacaine. After surgery, all catheters were injected with contrast media and examined radiographically. RESULTS The frequency of maldistribution was not significantly different in the isobaric and hyperbaric groups. A caudally oriented catheter tip was found to be a major cause of maldistribution (P < 10(-5)). A thoracic sensory level could be reached in all patients presenting a limited block by simply changing the baricity of the bupivacaine, the position of the patient, or both. The sensory level obtained 10 min after the first injection of 2.5 mg isobaric or hyperbaric bupivacaine was found to be a predictive factor of maldistribution. CONCLUSIONS Hyperbaric solutions do not appear to be a clinical factor in the development of limited block. The principle factor causing the maldistribution of bupivacaine is the caudal orientation of the tip of the end-hole catheter rather than its level or the route of catheter travel.