Two techniques to measure muscle fiber conduction velocity (MFCV) were compared. First, muscle fibers of biceps muscle were directly stimulated with needle electrodes, and the latency of the evoked muscle fiber action potentials was measured at a distance of 5 cm. Subsequently, the MFCV was measured at the same place with surface electrodes using the cross-correlation method. Fourteen controls were studied and illustrative results of the measurements of 6 myopathy patients are given. A clear correlation between the mean values of the two methods was found. The surface EMG technique resulted in a systematically higher MFCV (mean 1.0 m/sec); the variability of MFCV was much higher with the invasive technique. The reasons for these differences are discussed. MFCV measurements are shown to be of diagnostic value in some myopathies, for example myositis. In myopathies with a global reduction of the MFCV the two methods are of equal value; in some cases of longstanding myositis the needle method demonstrated some very slowly conducting fibers which were not detected with the surface method.