BACKGROUND Our aim was to determine the clinical efficacy of levobupivacaine for awake spinal anesthesia in ex-premature infants having lower abdominal surgery. The secondary aim was to determine the minimum local analgesic dose (MLAD) defined as the median effective local anesthetic dose for spinal anesthesia in neonates. METHODS Fifty neonates < 55 weeks postconceptional age having lower abdominal surgery were enrolled. The dose of levobupivacaine was determined by up-down sequential allocation. The first infant received 1.25 mg x kg(-1) 0.5% levobupivacaine with subsequent doses determined by the response of the previous patient. Dose interval was 0.25 mg x kg(-1). RESULTS There were no significant adverse effects attributable to levobupivacaine. Success rates for 1.25, 1, 0.75 and 0.5 mg x kg(-1) were 100, 83, 81 and 46%, respectively. Mean duration (95% confidence interval) of lower limb motor blockade was 81.7 min (75.8-87.6 min). Duration of anesthesia far exceeded surgical duration in all but one case. The up-down sequences were analyzed with the Dixon and Massey method. The motor block MLAD (95% confidence interval) was 0.48 mg x kg(-1) (0.42-0.54). Probit analysis was used to compare calculated ED50 and to calculate an ED95 by extrapolation of data into the clinical range. The ED(50) (95% C.I.) was 0.52 mg x kg(-1) (0.30-0.73). The calculated ED95 was 1.14 mg x kg(-1) (0.78-1.50). These values were confirmed by logit analysis. CONCLUSIONS Levobupivacaine is an effective agent for spinal anesthesia in neonates at a recommended dose of 1 mg x kg(-1). It appears to be of equivalent potency to racemic bupivacaine for motor blockade in ex-premature infants requiring lower abdominal surgery.