We compared the neurotoxic profile of racemic bupivacaine and levobupivacaine in: (i) a mouse model of N-methyl-D-aspartate (NMDA)-induced seizures and (ii) in an in vitro model of excitotoxic cell death. When used at high doses (36 mg/kg) both bupivacaine and levobupivacaine reduced the latency to NMDA-induced seizures and increased seizure severity. However, levobupivacaine-treated animals underwent less severe seizures as compared with bupivacaine-treated animals. Lower doses of levobupivacaine and bupivacaine had opposite effects on NMDA-induced seizures. At doses of 5 mg/kg, levobupivacaine increased the latency to partial seizures and prevented the occurrence of generalized seizures, whereas bupivacaine decreased the latency to partial seizures and did not influence the development of generalized seizures. In in vitro experiments, we exposed primary cultures of mouse cortical cells, containing both neurons and astrocytes, to 100 microM NMDA for 10 min for the induction of excitotoxic neuronal death. This treatment killed 70-80% of the neuronal population, as assessed 24 h after the excitotoxic pulse. In this particular model, both levobupivacaine and bupivacaine were neuroprotective against NMDA toxicity. However, neuroprotection by levobupivacaine was seen at lower concentrations (with respect to bupivacaine) and was maintained at concentrations of 3 mM, which are much higher than the plasma security threshold for the drug in vivo. In contrast, no protection against NMDA toxicity was detected when 3 mM concentrations of bupivacaine were applied to the cultures. Our data show a better neurotoxic profile of levobupivacaine as compared to racemic bupivacaine, and are indicative of a safer profile of levobupivacaine in clinical practice.