To refine the technique of phrenic nerve conduction, we first studied electrode positioning and sources of chest wall artifact. Diaphragmatic compound motor action potentials (DCMAPs) were mapped at close intervals over 4 hemithoraces of two subjects, finding optimum recording sites which were then used to quantitate artifacts due to EKG, chest wall EMG, and configurational thoracic changes of respiration. Based on these findings, 20 normal subjects were studied, showing ease of application and good side-to-side agreement for DCMAP latencies; but, in contrast to prior reports, right-to-left correlation for amplitude and waveform was poor, making the unaffected side an unreliable standard in unilateral partial phrenic nerve lesions.