The Diabetes Core Dataset is a minimum set of 30 variables considered essential for most diabetes research projects. It is designed to for implementation using routinely collected electronic health record data from primary care, with optional linkage to secondary care datasets such as Hospital Episode Statistics (HES). Variables were selected through expert consensus via the HDR UK Diabetes Data Science Catalyst (https://bhfdatasciencecentre.org/areas/diabetes-data-science-catalyst/) involving clinicians, researchers, and patient and public representatives. Core domains include demographics, diabetes characteristics, clinical measurements, biomarkers, diabetes-related complications, lifestyle factors and medications. It provides a standardised, reproducible framework that balances comprehensiveness with feasibility in routine NHS research data, enabling consistent analyses across studies while remaining adaptable to different data sources. It also offers a structured starting point for ingesting diabetes-related data into NHS Secure Data Environments (SDEs), enabling more efficient and standardised cross-site data preparation for research. The Diabetes Core Dataset differs from other datasets such as the National Diabetes Audit in that it has been developed specifically for research applications rather than service evaluation. It is designed as a minimum dataset that can be linked to specialist registries and external data sources to address more specific research questions.