Earlier literature documented high level of inaccuracies in the current cause-of-death reporting system for in-hospital deaths. Discrepancies are typically identified from two sources: comparing death certificates to autopsy results and/or to patient health (often, hospital) records. But many patients do not receive autopsy, and many die out of hospital. This paper examines the accuracy of the reporting of diabetes as a contributing cause of death. Using linked mortality records (2020) and Electronic Health Records (EHR) (2019-2020) from the state of Indiana, we overcome data limitations of the earlier research and also address such questions as whether diabetes reporting depends on the type of diabetes or patient characteristics, among others. Preliminary analysis shows that among persons with diabetes diagnosis in EHR (34%), the likelihood that death certificate refers to diabetes is low – 31%. When diabetes is referred to in death records, it is rarely the primary cause of death (5.2%); it is commonly listed as “complications of diabetes”. We next examine heterogeneity in diabetes reporting and proceed to identifying policy implications of its underreporting. Accurate information on cause of death and comorbidities is needed for reliable cause-specific mortality estimates, and to support allocation of healthcare resources and research effort.