The worldwide prevalence of dementia is predicted to rise significantly in the next three decades. However, these projections have not taken into account the role of modifiable risk factors and whether any prevention strategies might influence the predicted trend. Attempts at pharmacological disease modification have largely been disappointing, and the difficulties in conducting dementia trials are reviewed here. In contrast, recent population studies in high-income countries suggest that the epidemiology may be changing with a possible decline in incident dementia, or even a reduction in age-specific prevalence. Therefore, efforts to develop public health interventions may prove to be the more successful approach to addressing dementia at a societal level.