Walking is a fundamental movement skill in humans. However, how the brain controls walking is not fully understood. In this functional magnetic resonance imaging study, the rhythmic, bilaterally alternating ankle movements were used as paradigm to simulate walking. In addition to the resting state, several motor tasks with different speeds were tested. Independent component analysis was performed to detect four components shared by all task conditions and the resting state. According to the distributed brain regions, these independent components were the cerebellum, primary auditory cortex–secondary somatosensory cortex–inferior parietal cortex–presupplementary motor area, medial primary sensorimotor cortex–supplementary area–premotor cortex–superior parietal lobule, and lateral primary somatosensory cortex–superior parietal lobule–dorsal premotor cortex networks, which coordinated limb movements, controlled the rhythm, differentiated speed, and performed a function as a basic actor network, respectively. These brain networks may be used as biomarkers of the neural control of normal human walking and as targets for neural modulation to improve different aspects of walking, such as rhythm and speed.