Harlequin syndrome is a curious phenomenon in which one half of the face fails to flush during thermal or emotional stress as a result of damage to vasodilator sympathetic fibers. Anecdotal reports suggest that some of these patients have abnormal pupils. In this study we set out to systematically investigate autonomic pupil disturbances in an unselected cohort of patients with harlequin syndrome. A consecutive series of 39 patients with harlequin syndrome who were referred to a tertiary autonomic function laboratory underwent slit-lamp examinations, testing of deep tendon reflexes, infrared video pupillography and, where needed, additional pharmacologic pupillary testing. Results were compared with a meta-analysis of all previously reported cases of harlequin syndrome (n = 39) identified from a literature search. In 65% of patients, no underlying causative medical disturbance could be identified. In 64% of patients, there were abnormal pupils, most commonly Horner syndrome, which was always present ipsilateral to the side of the face with impaired facial sweating and flushing. The lesion was postganglionic in 9 of 10 patients tested pharmacologically. Five (13%) patients had tonic pupils, most of whom also had tendon areflexia but no other neurologic findings, a pattern consistent with Holmes-Adie syndrome. In 2 of these patients, tonic and Horner pupils coexisted. Normal pupils were present in 36% of patients. These results are similar to those for the 39 previously reported patients with harlequin syndrome. The frequent coexistence of harlequin and Horner syndromes without other neurologic deficits suggests pathologic changes affecting the superior cervical ganglion. Because either syndrome may occur alone, damage is apparently selective. Among the patients with harlequin syndrome who also have tonic pupils and tendon areflexia (Holmes-Adie syndrome), we postulate a ganglionopathy affecting not merely the (sympathetic) superior cervical ganglion, but also the (parasympathetic) ciliary and dorsal root ganglia. Because we found that more than 10% of patients had an undisclosed mass lesion in the chest or neck or a generalized autonomic neuropathy, we recommend a targeted evaluation in selected patients with harlequin syndrome.