Background: Current evidence suggests an inverse association between socioeconomic status and stroke incidence. Our aim was to measure the variation in incidence among different city districts (CD) and their association with socioeconomic variables. Methods: We prospectively ascertained all possible stroke cases occurring in the city of Joinville during the period 2005–2007. We determined the incidence for each of the 38 CD, age-adjusted to the population of Joinville. By linear regression analysis, we correlated incidence data with mean years of education (MYE) and mean income per month (MIPM). Results: Of the 1,734 stroke cases registered, 1,034 were first-ever strokes. In the study period, the crude incidence in Joinville was 69.5 per 100,000 (95% confidence interval, 65.3–73.9). The stroke incidence among CD ranged from 37.5 (22.2–64.6) to 151.0 per 100,000 (69.0–286.6). The stroke incidence was inversely correlated with years of education (r = –0.532; p < 0.001). MYE and MIPM were strongly related (R = 0.958), resulting in exclusion of MIPM by collinearity. Conclusions: Years of education can explain a wide incidence variation among CD. These results may be useful to guide the allocation of resources in primary prevention policies.