OBJECTIVE: Deficiency in antisaccade performance has been proposed as a schizophrenia endophenotype. METHOD: The authors assessed performance on an antisaccade task (and a prosaccade control condition) by 10 monozygotic twin pairs discordant for DSM-IV schizophrenia and 10 monozygotic healthy twin pairs matched for age, sex, and parental socioeconomic status. The authors computed antisaccade gain, latency, and error rate, as well as prosaccade gain and latency. RESULTS: The schizophrenic twins made more antisaccade reflexive errors than the nonschizophrenic co-twins and comparison twins, who did not significantly differ from each other. The nonschizophrenic members of discordant pairs performed worse than the comparison twins on antisaccade gain and latency but did not differ from their schizophrenic co-twins on these variables. There were no differences on prosaccade performance. Antisaccade errors were correlated with negative symptoms in the patients. CONCLUSIONS: Antisaccade spatial accuracy and latency deficits may serve as markers of genetic liability for schizophrenia.