Cluster headache is described here as having three distinct and contiguous clinical phases. Evidence of the pathophysiological changes associated with each phase is reviewed. The first phase, the cluster period, is characterized by chronobiological aberrations and impaired sympathetic nervous system activity. These changes may result in impaired autoregulatory chemoreceptor activity and susceptibility to attack provocation. An hypothesis that attempts to explain the second phase, cluster attack induction, is reviewed. Evidence for this model suggests that as a result of chemoreceptor dysfunction, a sustained hypoxemic event, as may result from altitude hypoxia, sleep apnea, or vasodilators, could provoke the cluster attack. Attack symptoms and signs, which constitute the third phase of cluster headache, are likely the result of parasympathetic and trigeminal nerve stimulation. Specifically, cluster headache pain is likely the consequence of neurovascular inflammation, as hypothesized in the trigeminovascular theory.