The ability to identify unaffected gene carriers within families may be crucial to the success of schizophrenia genetics studies. Data collected from three family samples (N = 365) demonstrated that poor antisaccade performance is an exceptionally promising indicator of liability for schizophrenia. A particular antisaccade task version provides large separations (5-6 sigma) between proband and normal groups. Poor antisaccade performance alone correctly identified 70% of patients in California, Utah, and Micronesia schizophrenia samples. Twenty-five to 50% of these patients' nonpsychotic first-degree relatives also had poor antisaccade performance, yielding risk ratios around 20:1 for simplex and 50:1 for multiplex schizophrenia families. Poor antisaccade performance is associated with dorsolateral prefrontal cortex pathology, suggesting that dysfunction of this circuitry also may predispose individuals to developing this disease.