Incident stroke has long been recognized to cause dementia shortly after the event. Patients who survive stroke without early-onset poststroke dementia (PSD) are at a high risk of developing dementia months to years after the initial stroke incident, which has generated enthusiasm for exploring treatments to prevent delayed-onset PSD in survivors of stroke. However, results from clinical trials completed in the past 10-15 years have been disappointing. In light of these results, the present Review revisits the mechanisms of both early-onset and delayed-onset PSD and proposes preventive strategies and directions for future clinical trials. Early-onset PSD results from a complex interplay between stroke lesion features and brain resilience, whereas delayed-onset PSD is associated mainly with the presence of severe sporadic small vessel disease (SVD), and to a lesser extent with Alzheimer disease pathology or recurrent stroke. As well as preventing stroke and delivering acute stroke treatments to reduce initial brain damage, measures to increase brain resilience could also reduce the risk of developing dementia if an incident stroke occurs. Future efforts to prevent delayed-onset PSD should focus on the study of sporadic SVD and on evaluating whether other strategies, in addition to conventional secondary stroke prevention, are effective in dementia prevention in this high-risk group.