Cerebral blood flow (CBF) was measured twice at years interval using xenon-133 inhalation technique in patients with atherothrombotic and embolic supratentorial brain infarction. The purpose of the study was to elucidate factors influencing long-term change in CBF in two subtypes of brain infarction of different mechanisms. Those patients were excluded from the study, who had bilateral hemispheric lesions, significant arterial lesion in the contralateral carotid axis and recurrent stroke before the second measurement of CBF. Of 46 patients studied, 23 (17 men and 6 women) were classified as atherothrombotic and 23 (15 men and 8 women) as embolic infarction based on the diagnostic criteria reported earlier. Their age at onset was 60.6 +/- 8.9 years old (mean +/- S.D.) for atherothrombotic patients, and 60.2 +/- 11.0 years old for embolic patients. The first measurement of CBF was performed between 31 and 87 days and the second measurement between 13 and 99 months after onset. In atherothrombotic group, mean hemispheric CBF (mCBF) of the affected side tended to be higher in patients examined at intervals shorter than 30 months, but tended to decrease in those at longer interval. The multivariate stepwise regression analysis indicated only "the interval between 2 measurements" to have a significant effect on reduction of mCBF, although in addition to the interval (p less than 0.05), infarct-size (p less than 0.01) and change in PaCO2 (p less than 0.05) were suggested to be possible factors by single regression analysis. In embolic group, an increase in hematocrit had a significant effect on reduction of mCBF, even when being evaluated with multivariate stepwise regression analysis.(ABSTRACT TRUNCATED AT 250 WORDS)