Although a few reports of phrenic nerve stimulation have appeared over the past dozen years, electromyographers have not readily accepted these methods. Perhaps this is because the techniques are considered unreliable, or because there is a major element of patient discomfort. In this study a simple, safe, and relatively painless procedure was devised for placing a needle near the phrenic nerve for use as a stimulating electrode. Also, the positions of the recording electrodes were relocated until a site was found that consistently provided a response with an initial negative deflection in all subjects. Emphasis has been on improving technique. A group of 60 nerves in 30 normal subjects was studied. The mean latency was 7.44 +/- 0.59msec, and the mean difference between right and left was 0.08 +/- 0.42msec. A response was obtained in all nerves examined. In addition, 13 patients with quadriplegia due to spinal cord injury were studied. This method now permits reliable analysis of localized phrenic nerve lesions and abnormalities induced in either the phrenic nerve or diaphragm by generalized disease processes.