An attempt was undertaken to assess the diagnostic value of sensory potentials evoked by stimulation of leg nerves. Findings in normal persons were as follows. First, stimulation of the sural nerve was superior to other methods, especially stimulation of the tibial nerve. Second, variations in latency were considerable and not attributable to age, height, or skin temperature. In many normal persons latency was not stable in either short-time or long-time trials. Amplitudes varied to such an extent that they could not help in diagnosis with the possible exception of extreme side-to-side differences. Finally, both amplitudes and latencies varied in relation to stimulus intensity, repetition rate, and filtering. Since sensory potentials from the leg nerves varied considerably, normal values must extend over a range that is wide enough to avoid mistaken diagnosis of abnormality. Hence, slight disturbances of nerve conduction such as those found in certain neuropathies, root damage and many extramedullary intraspinal space-occupying lesions could not be identified, whereas extensive demyelination in all parts of the sensory neuronal chain was readily discovered.