We explored the effect of education and occupational complexity on the rate of cognitive decline (as measured by the Mini-Mental State Examination) in 171 patients with a confirmed Alzheimer's disease (AD) diagnosis. Complexity was measured as substantive complexity of work and complexity of work with data, people, and things. Average lifetime occupational complexity was calculated based on years at each occupation. Participants were followed for an average of 2.5 years and 3.7 visits. In multivariate mixed-effects models, high education, high substantive complexity, and high complexity of work with data and people predicted faster rates of cognitive decline, controlling for age, gender, native language, dementia severity, and entry into the analyses at initial versus follow-up testing. These results provide support for the concept of cognitive reserve according to which greater reserve may postpone clinical onset of AD but also accelerate cognitive decline after the onset.