To explore cohort effects on age- and Alzheimer disease (AD)-related neuropathologic changes. 
We compared amyloid deposition in autopsied cases aged 65 years and older who died between 1972 and 2006. We included consecutive cases for 1972-1975, 1980, 1985, 1990, 1995, and 2000-2006
. We used linear regression models to assess period effects after adjustment for age, cognitive status, and neurofibrillary tangle (NFT) staging. We calculated amyloid/NFT stage ratios to account for possible changes in AD prevalence/severity over time. 
Mean amyloid stage was significantly related to year of death (p = 0.001) in the total population (1,599 cases, mean age 82 ± 8 years) and decreased 24%, from 1.88 ± 0.89 to 1.57 ± 0.81 (p < 0.0001), in 1,265 individuals without dementia. This decrease was particularly marked in the oldest age groups; people 85 years and older in 2006 had less amyloid deposition compared with those aged 75 to 84 years in 1972. Recent cohorts had lower amyloid deposition. The amyloid/NFT stage ratio decreased from 1.51 ± 0.74 to 0.99 ± 0.56 (p < 0.0001) in cases without dementia and from 0.74 ± 0.13 to 0.56 ± 0.21 (p = 0.0019) in individuals with dementia, confirming that more recent cases had less amyloid despite higher NFT densities
. Cohort effects were highly significant (p < 0.0001). The strong cohort effect we describe may influence the performance of early amyloid-based AD markers. It also provides preclinical evidence supporting recently described decreases in AD incidence.....