Amyloid-β (Aβ) pathology is a major component in the mechanisms behind Alzheimer's disease (AD). Measurement of Aβ(42) in cerebrospinal fluid predicts cognitive decline in patients with mild cognitive impairment and identifies AD in patients with dementia. However, studies on Aβ in plasma are contradictory. In this prospective population-based study, plasma Aβ(42) and Aβ(40) were measured at baseline in 730 adults aged 70 years or older and without dementia. After five years, plasma levels were analyzed again and participants were assessed for development of dementia. During follow-up, 53 individuals (7%) developed dementia of which 37 (5%) were classified as AD. No difference in baseline plasma Aβ(42), Aβ(40), or Aβ(42)/Aβ(40) ratio levels were observed between converters to dementia or AD compared to the cognitively stable individuals. However, individuals with plasma Aβ(40) levels above the median level for the group at baseline had an increased risk of developing dementia and AD during the follow-up, even after adjustment for age, gender, APOE genotype, and educational level (odds ratio = 2.2, 95% confidence interval = 1.0-4.7, p < 0.05). Neither plasma Aβ(42) nor the Aβ(42)/Aβ(40) ratio influenced the risk of developing dementia or AD. Moreover, Aβ(42) and Aβ(40) levels increased over the 5 years, whereas the Aβ(42)/Aβ(40) ratio decreased (p < 0.001). In conclusion, this study suggests that measurement of plasma Aβ should not be used clinically to predict dementia or AD. However, plasma Aβ(40) may possibly be regarded as a moderate risk marker comparable to other risk markers for AD such as first-degree family history of dementia.