Background: One of modifiable risk factors of diabetes is unhealthy diet which is related to obesity. Individuals with childhood obesity are at higher risk of adulthood obesity. School-based diabetes prevention programs are important to reduce childhood obesity. When resources are limited, evidence-based priority need to be undertaken. However, data related to childhood obesity was not widely available, the study used diabetes-related death records as the proxy of diabetes burden. This study aimed to map and identify geographic variation of diabetes-related mortality rate by school district level in Michigan to be used for policy-relevant information.Design and Methods: This study used death records in Michigan. Diabetes-related mortality rate and years potential life lost (YPLL) was calculated at the school district level. Spatial autocorrelation local Moran’s I and geographically weighted regression were used to evaluate spatial pattern of age-adjusted diabetes-related mortality rate by school districts.Results: The age-adjusted diabetes-related mortality rate ranged from 17.0 (95% CI, 8.6-25.5) to 171.3 (95% CI, 135.9-206.7) deaths per 100,000 population. The YPLL per person ranged from 0 to 19.3 years (95% CI, 15.5-23.1). High rates of diabetes-related mortality rate and YPLL clustered in East central and Southeastern region of Lower Peninsula Michigan including Flint, Kearsley, Beecher, Westwood Heights, Detroit, Ecorse, River Rouge, Taylor, Allen Park and Lincoln Consolidated school districts.Conclusions: There was variation in diabetes burden examined by diabetes-related mortality rate and YPLL at the school district level within Michigan State. The high cluster can be prioritized for the intervention programs.Significance for public healthChildren with obesity are at higher risk of having obesity during their adulthood. Because obesity is related to diabetes, having diabetes school-based prevention program that focused on reducing childhood obesity as well as diabetes is very beneficial. With limited resources, priority has to be made by the government based on diabetes burden in the area. To evaluate diabetes burden, the study used diabetes death records as the proxy because data related to diabetes risk factors among school children are not available at the school district level, and limited data of diabetes prevalence among youth at sub-county level. Our finding suggests the potential of using diabetes death record as the proxy to describe diabetes burden at the school district level. The results indicated the variation and clusters of diabetes in Michigan, which can be used for priority of the program. This approach also can be used in other public health program using available secondary data to provide policy-relevant information.