Objective: this study investigates the role of social and mental occupational characteristics in cognitive decline after retirement. Methods: the study included 1,048 subjects aged >/=65 years from the Three City cohort. Participants were evaluated at home at the initial visit and at 2-year intervals for a period of 12 years. The study includes detailed assessments of cognition, health and information about the subjects' main occupation. The four cognitive tests have been grouped into one latent factor. Three independent raters specialised in employment were asked to evaluate the level of social and intellectual stimulation for each occupation, which was then rated as low, medium and high. Results: after controlling for potential confounding factors, no association was found between higher levels of social stimulation at work and baseline cognition (medium score, P = 0.440; high score, P = 0.700) as compared with a low level. While cognitive trajectories were initially similar between high and medium levels of social stimulation compared with that of a low level, with advancing age this association diverged whereby more social stimulation during work years was related to accelerated cognitive decline that further grew in magnitude with older age. For mental stimulation, differences were only observed at baseline, with greater levels of mental stimulation during work years being associated with better cognitive performance (medium score, beta = 0.573, P = 0.015; and high score, beta = 0.510; P = 0.090) compared with a low level of mental stimulation. Conclusion: workers retiring from occupations characterised by high levels of social stimulation may be at risk of accelerated cognitive decline with advancing age.< Réduire