The reliability of the electrophysiological criterion of conduction block in determining the presence of focal demyelination in a peripheral nerve has been studied in an animal model. Demyelination was produced in the rat tibial nerve by one or two closely spaced microinjections of lysophosphatidylcholine (LPC). Histological and electrophysiological data were obtained on the acute lesion (up to 6 days), and during recovery (up to 11 weeks). Single LPC injections produced a lesion of very variable severity. Double injections more reliably produced a severe lesion with marked conduction block. Slight axonal damage was occasionally seen in nerves showing severe demyelination. The ratio of amplitude of muscle action potentials evoked by stimuli proximal and distal to the sites of nerve injection was calculated to detect the development of conduction block. The post injection ratio was more than 2 standard deviations below the control mean in 86% of nerves showing signs of demyelination. No control saline injected nerves showed such evidence of conduction block. The severity of the electrophysiological abnormality did not prove a reliable indicator of the severity of the histological lesion, however. The possible reasons for this variability are discussed and it is argued that caution should be exercised when interpreting this particular electrophysiological finding in clinical practice.