Normal elderly control participants showed short-term (10-min delay) and long-term (12 months delay) priming on the Gollin Figures Test. Nearly all patients with Alzheimer's disease or vascular dementia showed short-term priming, but the magnitude of their priming was less than that of controls. Significant long-term priming was not observed for the dementia groups. Differences between controls and dementia patients on the short-term priming test may depend upon structural-perceptual processes that are intact in dementia patients and controls and explicit memory functions available only to controls. The same model could account for differences between normal elderly and dementia patients on the long-term priming test, but several other explanations are also plausible.