We quantitated the size and configuration of compound muscle action potentials (CMAPs) in 266 nerves (66 median, 67 ulnar, 71 tibial and 62 peroneal) of 72 patients with acute inflammatory demyelinating polyneuropathy (AIDP) initially studied within 19 days of symptom onset. Results were compared with criteria for CMAP abnormalities, including criteria for abnormal negative peak duration and desynchronisation, derived from a control population of 50 median, ulnar, tibial and peroneal nerves. Other motor conduction abnormalities including minimal F response latency were also examined. We also analysed patterns of CMAP abnormality, peak disability and outcome for AIDP patients who had at least 3 motor nerves evaluated at first electrophysiologic study. Amongst AIDP nerves, low amplitude of the distal CMAP, usually with prolonged distal latency, was much more common than an abnormal fall in CMAP amplitude between stimulus sites. Using our CMAP criteria more than half of these low amplitude distal responses showed prolonged negative peak duration of desynchronisation or both, consistent with demyelination. Of the 47 AIDP patients who had 3 or more nerves initially studied, 37 (78.7%) had at least 1 motor nerve with a distal CMAP showing evidence of temporal dispersion. In addition, those with at least 75% of motor nerves showing a pattern of low amplitude of the distal CMAP without a further significant fall in amplitude between stimulus sites had greater peak disability and a poorer outcome. Assessment of temporal dispersion of the distal CMAP should be included in electrophysiologic criteria for acute demyelination. In addition, for some patients with AIDP patterns of CMAP amplitude abnormality amongst motor nerves are present early in the illness and may provide prognostic information.