SUMMARY Three hundred and fifty-two patients with atherosclerotic middle cerebral artery stenosis (MCAS, 53%) or occlusion (MCAO, 47%) have been systematically studied. The study involved all patients entered into the EC/IC Bypass Study with isolated MCA disease or a tandem lesion predominating in the MCA ipsilateral to the ischemic events (18 patients with a tandem lesion of greater magnitude in the internal carotid artery were not included). The Asian patients represented 58% of all Asians entered into the EC/IC Bypass Study, whereas the white patients represented 18% of all whites and the black patients 34% of all blacks. Isolated TIAs were less frequent in MCAO (12%) than in MCAS (34%). Warning TIAs before a stroke occurred in one third of the cases. Presentation with stroke or isolated TIA was not influenced by sex, age, level of MCA obstruction, collateral circulation nor associated carotid disease. In MCAS, no major difference in presentation was found between severe and moderate stenosis. Pure motor hemiparesis occurred in 15% and pure sensory stroke in 2% of the patients with stroke and 30% of the MCA territory infarcts were small and limited to the lentkulocapsular area, confirming that so-called lacunar infarcts may be due to large vessel disease.During follow-up (42 months) of 164 medically-treated patients, further cerebrovascular events (TIA and stroke) occurred in 11.7% of the patients per year. In MCAO the stroke rate was 10.1 % per patient-year and the ipsilateral infarct rate was 7.1% per patient-year. In MCAS, the stroke rate was 9.5% per patientyear and the ipsilateral stroke rate was 7.8% per patient-year. The location and severity of lesion did not influence the occurrence of ipsilateral ischemic events during follow-up. Reopening of an artery-to-artery embolic occlusion of the MCA, with subsequent embollc reocclusion, may explain some of the ipsilateral ischemic events during follow-up. The annual death rate was 3.3% in MCAS and 2.6% in MCAO. Less than 15% of the survivors were severely disabled at the end of follow-up and nearly two-thirds were able to resume previous activities both hi MCAO and MCAS. The type of delayed ischemic events tended to be the same as that of the presenting event, but no factor could significantly predict the occurrence of stroke or death. This study suggests that long-term prognosis of patients with MCA occlusion, who present with TIA or non-devastating stroke, is reasonable.