The purposes of this study were to quantify the effectiveness of various methods to minimize cross talk in surface electromyography (EMG) using common recording equipment, and to compare the intra- and interday variabilities of signals recorded with these methods. Comparisons were made for signals recorded with the single differential (SD), double differential (DD) and branched electrode (BE) techniques with large and small electrodes and corresponding interelectrode distances. The amount of cross talk in tibialis anterior EMG signals during maximum voluntary effort (MVE) triceps surae excitation was estimated using a protocol involving electrical stimulation of the triceps surae via the tibial nerve. In SD signals, cross talk averaged 12.2 and 10.2 percent MVE, for the large and small interelectrode distances, respectively. DD and BE signals contained significantly less cross talk (approximately 5 percent MVE for both techniques and interelectrode distances) than SD signals. The intra- and interday variabilities associated with these methods were estimated by recording tibialis anterior EMG signals during maximum voluntary isometric dorsiflexion (3 trials on each of 2 days) and calculating coefficients of variation for average-rectified values and median frequencies. EMG signals recorded with the small interelectrode distance showed greater interday amplitude variability than those recorded with the large interelectrode distance. Intra- and interday amplitude variabilities were similar across SD, DD and BE recording techniques. Intra- and interday frequency variabilities were similar across all experimental conditions. Thus, the DD and BE techniques, in conjunction with the large interelectrode distance (and large electrodes), provide a signal which contains significantly less cross talk than the SD technique without sacrificing intra- and interday amplitude and frequency stability.