Group means from nerve conduction studies of persons exposed to lead were used in a meta-analysis. Differences between the control and exposed groups, and the slopes between nerve conduction measurements and log(10) blood lead concentrations were estimated using mixed models. Conduction velocity was reduced in the median, ulnar, and radial nerves in the arm, and in the deep peroneal nerve in the leg. Distal latencies of the median, ulnar, and deep peroneal nerves were longer. No changes in the amplitudes of compound muscle or nerve action potentials were detected. The lowest concentration at which a relationship with blood lead could be detected was 33.0 microg/dl for the nerve conduction velocity of the median sensory nerve. Lead may reduce nerve conduction velocity by acting directly on peripheral nerves or by acting indirectly, for example, on the kidney or liver.