It was unclear which type of recurrent stroke was linked to pulse pressure (PP). Therefore, the 1,454 participants in the PICASSO (Prevention of Cardiovascular Events in Ischemic Stroke Patients With High Risk of Cerebral Hemorrhage) database was researched. According to the distribution of mean PP (mm Hg) during follow-up (mean, 1.9 years), subjects were divided into quartiles: less than 47 (first quartile), 48 to 53 (second quartile), 54 to 59 (third quartile), and more than or equal to 60 mm Hg (fourth quartile). Hemorrhagic stroke was the primary endpoint, with ischemic stroke, a stroke of any sort, and significant adverse cardiovascular events as additional endpoints. The prediction accuracy of mean PP was assessed using an adjusted time-dependent area under the receiver operating characteristic curve analysis. The average number of blood pressure checkups was between 9.4 ± 5.5 times. In the increasing quartile of mean PP, the stroke incidence rate per 100 person-years was 3.14, 2.24, 5.52, and 6.22, respectively, while the central adverse cardiovascular events rate was 3.82, 2.84 6.37, and 7.14, respectively. Hemorrhagic stroke risk was higher in the highest quartile (adjusted hazard ratio, 6.03; 95% CI, 1.04–34.99) than the lowest quartile in the presence of mean arterial pressure, which was visible at higher mean systolic blood pressure. A higher mean PP likewise predicted hemorrhagic stroke as a continuous variable (1.09, 1.03–1.15). For hemorrhagic stroke, the time-dependent area under the receiver operating characteristic curve was 0.79. In stroke patients with cerebral microbleeds or intracerebral hemorrhage, long-term elevated PP combined with higher systolic blood pressure increased the risk of subsequent hemorrhagic stroke.