Patients with schizophrenia consistently perform worse than healthy controls on the antisaccade task in which the subject is required to inhibit a reflexive saccade to a suddenly appearing visual target and look in the opposite direction. To our knowledge there is no research yet showing how patients at ultra high risk (UHR) for developing psychosis perform on the antisaccade task. The aim of the present study was to investigate antisaccade task performance in UHR patients. Patients were eligible for the study when they met criteria for one or more of the following groups: Attenuated symptoms or brief limited intermitted psychotic symptoms or a first-degree family member with a psychotic disorder and reduced functioning or basic symptoms. In 35 UHR patients we assessed antisaccades, neuropsychological test performance and symptomatology. Antisaccade task results were compared with those obtained in 42 age- and intelligence-matched patients with recent-onset schizophrenia and 28 matched healthy controls. Antisaccade error rate was significantly higher in the UHR patients than in the controls. Schizophrenia patients performed worse than the UHR patients and the control subjects. We found a trend towards higher antisaccade error rate at baseline in the UHR patients who later made the transition to psychosis compared to the UHR patients who did not make the transition to psychosis. Poor spatial working memory function was related to increased antisaccade errors in the UHR group. Abnormal antisaccade task performance is also present in patients at UHR for developing psychosis. Subsequent research needs to clarify if increased antisaccade error rate is predictive of a psychotic episode. In UHR patients, poor antisaccade performance may reflect working memory dysfunction.