Objective: Our purpose was to assess whether family history of diabetes or the HLA-DR-DQ genotype of the index case was associated with the frequency of diabetic ketoacidosis (DKA) at diagnosis of childhood type 1 diabetes. Patients and methods: The study cohort comprised 1518 children aged !15 years and diagnosed with type 1 diabetes in Finland in [2002][2003][2004][2005]. Family history of type 1 and type 2 diabetes among firstdegree relatives (FDRs) and grandparents was assessed at diagnosis. HLA-DR-DQ genotypes were analysed using time-resolved fluorometry. Results: In total, 12.6 and 1.7% of children had at least one FDR affected with type 1 or type 2 diabetes, respectively, and 6.6 and 34.8% had at least one grandparent with type 1 or type 2 diabetes. DKA (pH !7.30) occurred less frequently in children having a type 1 diabetes affected FDR (7.4 vs 20.5%, P!0.001). Type 2 diabetes among the parents or grandparents had no such effect. Lower risk HLA genotypes were observed to predispose to DKA (P!0.024). In a logistic regression analysis, the risk of DKA was independently associated with the absence of a family member affected by type 1 diabetes, the presence of a low-risk HLA genotype and older age at diagnosis (odds ratio 3.23, 1.45 and 1.07 respectively). Conclusion: The presence of type 1 diabetes in an FDR is associated with an decreased risk of DKA at diagnosis. The rate of DKA seems to be higher in children with lower HLA-conferred risk for type 1 diabetes.