Background: Taking an advantage of the high sensitivity of 3D T2*-weighted gradient-recalled-echo (GRE) imaging to cerebral microbleeds, we investigated the relationship between cerebral microbleeds and leukoaraiosis. Methods: Participants aged 40 years or more have been evaluated for the presence of cerebral microbleeds using 3D T2*-GRE sequence since 2006. The severity of periventricular hyperintensity (PVH) and deep white matter hyperintensity (DWMH) on fluid attenuated inversion recovery images was assessed using Fazekas rating scales. Multivariate logistic regression analyses were conducted after adjustment for stroke subtype, age, PVH, DWMH, hypertension, dementia, and use of platelet aggregation inhibitors. Additionally, we examined the association between cerebral microbleeds and other covariates using a Pearson's correlation analysis. Results: Amongst 389 patients, 67 patients had a single microbleed and 93 had multiple microbleeds. The prevalence of microbleeds was 83% amongst 53 patients with intracerebral hemorrhage (ICH), 49% amongst 173 with infarction, and 20% amongst 163 without any type of stroke. In the multivariate analyses, the odds ratio (95% CIs) of microbleed detection was 10.1, (4.12-24.8) for ICH, 2.33 (1.12-4.85) for atherosclerotic infarction, 1.66 (1.10-2.48) for PVH, and 1.49 (1.02-2.19) for DWMH. In the Pearson ' s correlation analysis, cerebral microbleeds were closely related to PVH (Pearson's correlation coefficient; 0.48) and DWMH (0.37), compared with age (0.16). Conclusions: High-grade PVH, high-grade DWMH, ICH, and atherosclerotic infarction were significantly independent predictors for cerebral microbleeds. In addition, we found that the grades of PVH and DWMH have a closer association with the number of cerebral microbleeds than age.