Previous studies have shown that children with stress hyperglycemia have an increased risk for development of type I or insulin-dependent diabetes mellitus. To determine whether stress hyperglycemia in prospectively screened pediatric patients represents a prediabetic state. Prospective, cohort analytic study. The Children's Hospital of the King's Daughters is an urban pediatric emergency department at a tertiary care, university-based children's hospital in Norfolk, Va. All patients who required a venipuncture for evaluation of an acute illness or injury from October 1992 through March 1993 were screened prospectively for hyperglycemia (blood glucose level > or = 8.3 mmol/L [> or = 150 mg/dL]). Each hyperglycemic patient was age matched to a stress control subject (defined as a nonhyperglycemic but acutely ill child) from the emergency department and a healthy control subject from a well-child clinic. Blood samples were obtained at the time of initial evaluation in the emergency department from 30 hyperglycemic patients (age range, 4 weeks to 12.4 years; median, 2 years), 30 stress control subjects, and 30 healthy control subjects. All samples were tested for islet cell antibodies, insulin autoantibodies, glutamic acid decarboxylase (GAD) antibodies, and HLA typing, specifically the genotypes at the DQB1 gene. The presence of immunologic or genetic markers for insulin-dependent diabetes mellitus and/or the clinical development of insulin-dependent diabetes mellitus. No patients or control subjects were positive for islet cell antibodies. One hyperglycemic patient and 3 stress control subjects were positive for insulin autoantibodies; all 4 of these subjects had sickle-cell disease and fever. Four of the 8 patients with sickle-cell disease had insulin autoantibodies, compared with none of the 52 patients and stress control subjects without sickle-cell disease (P < .001). One healthy control subject had antibodies to GAD65. The patient group did not show increased genotypes at the DQB1 gene that were indicative of an enhanced risk for insulin-dependent diabetes mellitus. Of the 32 hyperglycemic patients, 27 healthy control subjects, and 25 stress control subjects contacted for follow-up at 31 to 36 months, none has developed insulin-dependent diabetes mellitus. Children with stress hyperglycemia do not have an increased prevalence of immunologic or genetic markers of insulin-dependent diabetes mellitus and thus do not appear to be at an increased risk for development of insulin-dependent diabetes mellitus. Our data suggest that insulin autoantibodies develop in children subject to sickle cell crises.