The purpose of this study was to determine the walking speed which has the greatest influence on neural relaxation in healthy elderly women as determined by electromyogram (EMG) and electroencephalogram (EEG) analyses. Seven elderly female volunteers [mean age 68.5 (SD 3.95) years] served as subjects for this study. The EMG signals were recorded from the gastrocnemius (MG), soleus (SL) and tibialis anterior (TA) muscles while walking on a treadmill, starting at 40 m.min-1 and increasing 6 m.min-1 incrementally for 10 min. The turning point of muscle activities (by integrated EMG. iEMGtp) was determined as the walking speed at the point at which the mean rate of change of iEMG (MG + SL + TA) abruptly increased. After the determination of iEMGtp. the treadmill was set at three constant speeds, one corresponding to the speed for the iEMGtp and two others 20% higher or lower than that for the iEMGtp. The subjects then walked for 20 min at each of these speeds on 3 separate days and their EEG power spectrum data were obtained for frequencies from the 8 to 13 Hz (z-wave component, AWC). The mean of iEMGtp for our subjects was at a mean walking speed of 64.7 (SD 7.9) m.min-1. Considering the subjects' age and height, iEMGtp was somewhat faster than their expected self-paced normal walking speed. There were no differences between the mean AWC values of the subjects prior to exercising at each of the three speeds. The mean AWC values after exercise were significantly (P < 0.01) greater than before. The extent of the increase in AWC at iEMGtp was greater than those at slower speeds. Our data would suggest that walking exercise at the speed which corresponds with EMG evidence of iEMGtp may induce the most significant relaxing effects in elderly women.