目的　评价脑梗死急性期OCSP分型的观察者间信度。方法　 2名神经科医生分别将连续就医、首次发生的急性脑梗死 1 1 0例分为 4个OCSP亚型。结果　 2名神经科医生分型的一致性为尚好 (kappa =0 3 9,95 %CI=0 3 0～ 0 48)至中度 (kappa =0 42 ,95 %CI=0 3 3～ 0 5 1 )。而对某些神经体征检查的一致性较差。结论　OCSP分型在脑梗死急性期的观察者间信度较为满意 ,是一种简便、实用的临床分型工具 ,神经体征变异是影响分型一致性的主要因素 Objective To evaluate the interobserver reliability of OCSP classification in acute cerebral infarction. Methods Two neurologists separately treated one hundred and ten cases of acute cerebral infarction for the first time and divided them into four OCSP subtypes. Results The consistency of the two neurologists was good (kappa = 0 39, 95% CI = 0 30 ~ 48) to moderate (kappa = 0 42, 95% CI = 0 3 3 ~ 0 5 1 ). The consistency of some neurological examination is poor. Conclusion OCSP classification is more satisfactory in the interobserver reliability of acute cerebral infarction and is a simple and practical tool for clinical typing. The variation of neurological signs is the main factor affecting the consistency of classification