INTRODUCTION Hypoglycemia is a defining feature of severe malaria and several other infectious diseases in children but the prevalence, significance, and prognosis of abnormal blood glucose, including hyperglycemia, have rarely been addressed in severely ill children in non-malaria endemic areas. METHODS In Madagascar, consecutive children (1 month-15 years) admitted to the pediatric ward of a referral hospital, were categorized using the integrated management of childhood illness (IMCI). Samples were taken once on admission for measuring blood glucose concentration. Glycemia levels (hypoglycemia <2.2 mmol/l; low glycemia: 2.2-4.4 mmol/l; normoglycemia >4.4-8.3 mmol/l; and hyperglycemia >8.3 mmol/l) were related to the IMCI algorithm and case fatality. Factors associated with blood glucose concentration and case fatality were analysed using univariate and multivariate analysis. RESULTS Of 420 children, 48.1% (n = 202) were severely ill; 3.1% (n = 13) had hypoglycemia; 20.0% (n = 84) low glycemia; 65.9% (n = 277) normoglycemia; and 10.9% (n = 46) hyperglycemia. In univariate analysis, hypoglycemia and hyperglycemia both showed significant increase in the risk of death, as compared to normal blood glucose (RR: 12.2, 95% CI: 6.2-23.7 and RR: 2.5, 95% CI: 1.0-6.2, respectively). Children with low glycemia had no increased risk of death (RR: 1.2, 95% CI: 0.4-3.2) despite a poorer IMCI status on admission. After logistic regression, hypoglycemia (RR: 19.4, 95% CI: 5.0-.74.7, hepatomegaly (RR: 12.2, 95% CI: 3.3-44.9) and coma (RR: 4.8, 95% CI: 1.3-17.6) were the features on admission associated with an increased risk of death. CONCLUSIONS Dysglycemia in non-neonates is associated with increased mortality. These findings underline the need for the use of rapid screening tests to initiate early treatment. Alternative treatments such as oral or sublingual administration of glucose should be developed in structures with limited resources.