Chronic electrical stimulation of the brain, known as deep brain stimulation (DBS), has become a preferred surgical treatment for medication-refractory movement disorders. Despite its remarkable clinical success, the therapeutic mechanisms of DBS are still not completely understood, limiting opportunities to improve treatment efficacy and simplify selection of stimulation parameters. This review addresses three questions essential to understanding the mechanisms of DBS. 1) How does DBS affect neuronal tissue in the vicinity of the active electrode or electrodes? 2) How do these changes translate into therapeutic benefit on motor symptoms? 3) How do these effects depend on the particular site of stimulation? Early hypotheses proposed that stimulation inhibited neuronal activity at the site of stimulation, mimicking the outcome of ablative surgeries. Recent studies have challenged that view, suggesting that although somatic activity near the DBS electrode may exhibit substantial inhibition or complex modulation patterns, the output from the stimulated nucleus follows the DBS pulse train by direct axonal excitation. The intrinsic activity is thus replaced by high-frequency activity that is time-locked to the stimulus and more regular in pattern. These changes in firing pattern are thought to prevent transmission of pathologic bursting and oscillatory activity, resulting in the reduction of disease symptoms through compensatory processing of sensorimotor information. Although promising, this theory does not entirely explain why DBS improves motor symptoms at different latencies. Understanding these processes on a physiological level will be critically important if we are to reach the full potential of this powerful tool.