Background: Most studies believed that lesion sites are decisive to the attack and types of aphasia, which is also in disputation. Objective: T o classify and evaluate aphasia with CT and MRI examinations, so as to reveal the association between aphasia types and lesion sites. Design: A cross-sectional study. Setting: Department of Neurology, Beij ing Tiantan Hospital affiliated to Capital University of Medical Sciences. Participants: The lesion sit es were selected from 1 198 patients with cerebral infarction, who were hospitalized in the stroke unit of the Department of Neurology, Beijing Tiantan Hospital affiliated to Capital University of Medical Sciences between February 2002 and February 2005, and totally 325 patients with aphasia were enrolled. Inclusive criteria: 1 the diagnosis accorded with the diagnostic standard of cerebral infarction set by National Meeting for Cerebrovascular Disease, and the patients with cerebral infarction were caused by lesions of left cerebral hemisphere proved by cranial CT or MRI; 2 the native language is Chinese; 3 those could cooperate and had consciousness; 4 the educational level as primary school and above, the intellect was normal before attack, and there was no history of mental disease; 5 primary attack or attack for many times without language disorders; 6 without severe liver and kidney diseases, and other diseases of internal medicine and surgery; 7 without other disease that could affect language function; 8 without other disease that could affect cognitive function; 9 the score of Western battery aphasia was ≤93.8. Exclusive criteria: 1 patients with cerebral infarction caused by lesion of right cerebral hemisphere; 2 the native language is not Chinese or mandarin is not standard; 3 illiterate; 4 attack for many times and with language disorders; 5 with memory and intellectual disorders; 6 with severe visual and auditory disorders; 7 uncooperative patients; 8 the score of Western battery aphasia was > 93.8. Totally 221 males and 104 females were enrolled, and the average age was (68.72±4.56) years. Methods: 1 The patients received CT examination before admission, and received MRI examination within 1 week after admission. The data collection and treatment were finished with the Siemens Trio 2003T magnetic resonance apparatus and Siemens AG 2003 work station. 2 The aphasia of the patients was classified and evaluated with the Western battery aphasia by professional language therapist within 2 weeks after admission. 3 The severity of aphasia was classified with the Boston diagnostic aphasia examination severity grading standard by the same language therapist on the same day of Western battery aphasia. There were 6 grades, grade 0 was taken as meaningless language or auditory understanding ability, and grade 5 as extremely few differentiable language disorders, the patients could feel some difficulties subjectively, but the hearer was uncertain to obviously detect. Main outcome measures: Association between aphasia types and lesion sites; Grading of aphasia severity. Results: All the 325 patients were involved in the analysis of results. 1 Aphasia not caused by the involvement of language centers: Of the 1 198 patients with cerebral infarction, the Broca area in 5 cases and Wernicke area in 4 cases were involved, and did not cause aphasia syndrome in the patients. 2 Aphasia types: Of the 325 patients with aphasia, the results of Western battery aphasia showed that 83 cases had Broca aphasia, 48 cases Wernicke aphasia, 58 cases complete aphasia, 12 cases conduction aphasia, 36 cases transcortical motor aphasia, 17 cases transcortical sensory aphasia, 19 cases transcortical mixed aphasia and 52 cases nominal aphasia. The lesions located at typical language center in 240 cases and at non-language centers in 85 cases. 3 Evaluative results of Boston diagnostic aphasia examination severity grading standard: It was grade 0 in 84 cases, grade 1 in 79 cases, grade 2 in 77 cases, grade 3 in 63 cases and grade 4 in 22 cases, and the typical language centers were involved in most of the patients of grade 0 and grade 1. Conclusion: The association between types and lesion sites of most aphasia are in accordance with the typical aphasia mode, but it was not completely accordant in a few aphasias, the lesion of non-language centers can also cause aphasia, and the aphasia is greatly severe in the patients with the lesion site of language centers.