The purpose of the study was to compare electrical stimulation (ES) and cervical magnetic stimulation (CMS) of the phrenic nerves for the measurement of the diaphragm compound muscle action potential (CMAP) and phrenic nerve conduction time. A specially designed esophageal catheter with three pairs of electrodes was used, with control of electrode positioning in 10 normal subjects. Pair A and pair B were close to the diaphragm (pair A lower than pair B); pair C was positioned 10 cm above the diaphragm to detect the electromyogram from extradiaphragmatic muscles. Electromyograms were also recorded from upper and lower chest wall surface electrodes. The shape of the CMAP measured with CMS (CMS-CMAP) usually differed from that of the CMAP measured with ES (ES-CMAP). Moreover, the latency of the CMS-CMAP from pair B (5.3 +/- 0.4 ms) was significantly shorter than that from pair A (7.1 +/- 0.7 ms). The amplitude of the CMS-CMAP (1.00 +/- 0.15 mV) was much higher than that of ES-CMAP (0.26 +/- 0.15 mV) when recorded from pair C. Good-quality CMS-CMAPs could be recorded in some subjects from an electrode positioned very low in the esophagus. The differences between ES-CMAP and CMS-CMAP recorded either from esophageal or chest wall electrodes make CMS unreliable for the measurement of phrenic nerve conduction time.