Longitudinal measurements of cognitive ability measured by serial testing using the Cognitive Capacity Screening Examination (CCSE) were correlated with cerebral blood flow (CBF) throughout (mean +/- SD) 19.9 +/- 12.6 months among 57 patients with multi-infarct dementia, 17 with dementia of the Alzheimer's type, 10 with both, and among 32 age-matched elderly normal controls. Longitudinal CCSE and CBF measurements among controls yielded stable normative values. Reduced mean CCSE scores correlated directly with CBF reductions in patients with multi-infarct dementia (p less than 0.0005) and dementia of the Alzheimer's type (p less than 0.028). Patients with multi-infarct dementia had CCSE scores with retest variability exceeding those of controls (p less than 0.001) and of patients with dementia of the Alzheimer's type (p less than 0.003). CCSE scores and CBF changed together 78.6% (p less than 0.001) of the time in patients with multi-infarct dementia compared with 66.2% of the time (p less than 0.01) in those with both, 62.9% of the time (p less than 0.05) in those with dementia of the Alzheimer's type, and 47.7% of the time (NS) in controls. Further analyses indicated that changes in CCSE scores and CBF were predominantly progressive declines in patients with dementia of the Alzheimer's type, whereas the changes were more bidirectional (both increases and decreases) in patients with multi-infarct dementia; these differences were also significant. Results support the diagnostic usefulness of the Hachinski ischemic scale and confirm that both cognition and CBF fluctuate together among patients with multi-infarct dementia, whereas patients with dementia of the Alzheimer's type exhibit a more stable course, with progressive declines in cognition and CBF.