Few studies have examined why, in most reports, low education is related to accelerated cognitive decline in middle and old age. Intellectual abilities - whether or not innate - and the mental stimulation provided by the educational process itself are frequently hypothesized to be the most relevant mechanisms. Work characteristics, such as the mental workload of a job, may be another mechanism, as these may also be related to educational level. Moreover, these are more amenable to modification than intellectual abilities. First longitudinal data from the Maastricht Aging Study (MAAS) among 708 men and women aged 50 to 80 in 1993-1995 were used to quantify the contribution of adult mental workload to the associationbetween educational level and age-related cognitive decline. In the 3-year follow-up, persons with a low educational level experienced more decline in cognitive function (information processing speed, memory, and general cognitive function according to the Mini-Mental State Examination) compared to persons with a high educational level. The low prevalence of mental stimuli and challenges at work among the poorly educated subjects explained about 42% of this association. The contribution was independent of crystallized intellectual abilities and was similar across measures of cognitive function. Our findings indicate that the gap in the risk of age-related cognitive decline between the poorly and highly educated persons may be substantially narrowed by increasing work-related mental stimuli and challenges among the poorly educated su bjects. Copyright 2003 Taylor & Francis Group, LLC. © 2003, Taylor & Francis Group, LLC. All rights reserved.