Inequalities in health persist at higher ages and might also affect cognitive functioning. E.g., it is a common finding that persons with higher educational achievement show higher levels of cognitive functioning and a lower risk of dementia. In contrast however, longitudinal studies altogether do not support the idea that higher educational achievement is protective against cognitive decline. One reason for this seeming discrepancy might result from the assumption that decline is either faster or slower for more educated persons, or occurs at the same rate for persons with different education. The pattern might however be more complex, with cognitive reserve hypothesis predicting that decline is initially slower but accelerates later on among more educated persons, while it is initially faster and might level off in the longer run among less educated persons. This hints towards the existence of differences in the shape of cognitive decline and calls for the study of relatively long age ranges, such that different shapes can become visible. Using growth curve models and data from the Longitudinal Aging Study Amsterdam, we examine decline in multiple cognitive domains after age 65 over a period of 20 to 25 years and investigate whether the initial rate of decline is slower and acceleration of decline is stronger for more (vs. less) educated persons. Studying such qualitative differences in the shape of cognitive decline with a unique dataset, our study will give important insights into more complex patterns of health inequalities at higher ages.