Lead (Pb2+) is a potent neurotoxin that acts as a non-competitive, voltage-independent antagonist of the NMDA receptor (NR) channel. Pb2+ action partially overlaps with that of zinc (Zn2+), but precise coincidence with Zn2+ binding site is debated. We investigated the site of Pb2+ interaction in NR channels expressed in Xenopus laevis oocytes from the clones zeta1, epsilon1 or epsilon2 and mutated epsilon1 or epsilon2 forms. For each epsilon subunit we chose two mutations that have been identified as 'strong mutations' for Zn2+ binding and examined the effect of Pb2+ on channels that contained those mutations. In epsilon1-containing channels, mutations D102A and H128A caused a decrease of Pb2+ inhibition with a 10-fold (D102A) and four-fold (H128A) shift of IC50. In epsilon2-containing channels, the most effective mutation in removing Pb2+ inhibition was H127A, with a five-fold increase of IC50, while D101A was virtually ineffective. Other mutations, D104A, T103A, and T233A, were less effective. The double mutation D101AH127A, while reducing Zn2+ inhibition by nearly nine-fold, caused a minor (less than two-fold) shift in Pb2+ IC50. Competition experiments showed that increasing doses of Zn2+ reduced the apparent affinity for Pb2+ in epsilon1-containing receptors, but not in epsilon2-containing receptors. In addition the effect of Pb2+ on epsilon2-containing channels was additive with that of ifenprodil, with no competition for the site. Although none of the mutations that we have tested abolished the block by Pb2+, our results indicate that the action of this toxic metal on NR channels is more dependent on the receptor composition than previously thought, because Zn2+ is able to displace Pb2+ from its binding site in epsilon1-containing channels, but not in epsilon2-containing channels.