To determine the optimal recording site for phrenic nerve conduction studies, six different recording techniques were compared in 11 healthy volunteers (22 phrenic nerves). The mean diaphragm compound muscle action potential (CMAP) amplitude, side-to-side difference, and the number of studies with a false-positive result (CMAP amplitude <0.30 mV) were compared for each technique. The largest amplitude (0.65 +/- 0.23 mV, range 0.30-1.2 mV) with good right-left agreement (mean difference 0.15 mV) and no false positives was obtained using technique 1, where the G1 electrode was positioned 5 cm above the xiphoid process and G2 16 cm from G1 along the costal margin. This was also the easiest technique to perform. It does not require rib counting, which may be difficult and inaccurate, especially in overweight patients. At least one false positive occurred with each of the remaining five techniques.