We compared the hemodynamic stability after spinal anesthesia with two different dosing regimens in the elderly. Fifty patients, all older than 60 yr and scheduled for elective knee or hip surgery were assigned to two groups. After administration of 10 mL/kg of lactated Ringer's solution (RL) intravenously (IV) in the first group, we performed a continuous spinal anesthesia (CSA) by means of a 28-gauge catheter through which repetitive injections of 2.5–5 mg of plain bupivacaine 0.5% were given. In the other group, a single-dose spinal anesthesia (SS) with 20 mg of the same local anesthetic (LA) was carried out. Noninvasive mean arterial pressure (MAP), heart rate, and levels of analgesia were measured. To maintain MAP within 25% of initial value, the patients received additional IV fluids (RL) as first measure. When MAP could not be maintained despite hydration, incremental doses of ephedrine were given IV. Six patients in the CSA group and 17 in the SS group developed a level of anesthesia higher than T6 (P < 0.01). In the SS group more fluid was needed (792 vs 388 ml) than in the CSA group (P < 0.01). Moreover, more patients of the SS group (11 vs 4) required ephedrine (P < 0.05). We conclude that CSA produces reliable and predictable analgesia for lower limb surgery with less need for correction of hemodynamic changes compared to SS.