In this study we investigated the association between plasma levels of eight forms of vitamin E and incidence of Alzheimer's disease (AD) among oldest-old individuals in a population-based setting. A dementia-free sample of 232 subjects aged 80+ years, derived from the Kungsholmen Project, was followed-up to 6 years to detect incident AD. Plasma levels of vitamin E (alpha-, beta-, gamma, and delta-tocopherol; alpha-, beta-, gamma-, and delta-tocotrienol) were measured at baseline. Vitamin E forms-AD association was analyzed with Cox proportional hazard model after adjustment for several potential confounders. Subjects with plasma levels of total tocopherols, total tocotrienols, or total vitamin E in the highest tertile had a reduced risk of developing AD in comparison to persons in the lowest tertile. Multi-adjusted hazard ratios (HRs) and 95% confidence interval (CI) were 0.55 (0.32-0.94) for total tocopherols, 0.46 (0.23-0.92) for total tocotrienols, and 0.55 (0.32-0.94) for total vitamin E. When considering each vitamin E form, the risk of developing AD was reduced only in association with high plasma levels of beta-tocopherol (HR: 0.62, 95% CI 0.39-0.99), whereas alpha-tocopherol, alpha- tocotrienol, and beta-tocotrienol showed only a marginally significant effect in the multiadjusted model [HR (95% CI): alpha-tocopherol: 0.72 (0.48-1.09); alpha-tocotrienol: 0.70 (0.44-1.11); beta-tocotrienol: 0.69 (0.45-1.06)]. In conclusion, high plasma levels of vitamin E are associated with a reduced risk of AD in advanced age. The neuroprotective effect of vitamin E seems to be related to the combination of different forms, rather than to alpha-tocopherol alone, whose efficacy in interventions against AD is currently debated.