Development of type 1 diabetes mellitus (T1D) may be triggered pre- or perinatally by multiple factors. Identifying new predisposing T1D markers or combinations of markers in a large, well-characterised case-control collection may be important for future T1D prevention. The present work describes the design and feasibility of a large and unselected case-control study, which will define and evaluate prediction criteria for T1D at the time of birth. Danish registries (Biological Specimen Bank for Neonatal Screening, and the National Discharge Registry) made it possible to identify and collect dried blood spots (DBS) from newborns who later developed T1D (cases) born 1981-2002. DBS samples from 2086 cases and two matching control subjects per case were analysed for genetic and immune factors that are associated with T1D: (a) candidate genes (HLA, INS and CTLA4), (b) cytokines and inflammatory markers, (c) islet auto-antibodies (GAD65A, IA-2A). The objective of the study was to define reliable prediction tools for T1D using samples available at the time of birth. In a unique approach, the study linked a large unselected and population-based sample resource to well-ascertained clinical databases and advanced technology. It combined genetic, immunological and demographic data to develop prediction algorithms. It also provided a resource for future studies in which new genetic markers can be included as they are identified.