Objective To investigate the relationship between cognitive function and eventrelated potential P300 in male long-term hospitalized patients with schizophrenia. Method Totals of 82 male long-term hospitalized patients with schizophrenia and 52 healthy controls were recruited. The eventrelated potential P300 was analyzed by the United States Nicolet Viking Quest evoked potential instrument. The cognitive function was tested by animal naming test， category fluency test， digital cancellation test， trail taking test (TMT-A， TMT-B)， stroop test (word， colour， word-colour interference test)， block design test and WMS-Ⅲ spatial span test. Results There were significant differences between patient group and control group in all items of cognitive function assessment scales (P＜ 0.01) . There were longer latency of P300 (t=22.990， P ＜ 0.01) and lower amplitude of P300 (t=-9.699， P ＜ 0.01) in patient group than that in control group. The P300 latency was positively correlated with digital cancellation test and the score of TMT-A in patient group (r=0.481，P＜0.01； r=0.245，P＜0.05). Moreover， P300 amplitude was negatively correlated with digital cancellation test in patient group (r= -0.338， P＜0.01). Conclusion The male long-term hospitalized patients with schizophrenia still have cognitive impairment， even when their psychiatric symptoms are stable. The latency and amplitude of event-related potential P300 may be an electrophysiological marker of cognitive function in schizophrenic patients， which were correlated with the results in the cognitive assessment scales.