Objective：To explore the effect of glycemic variability（GV） on the prognosis of critically ill children and to find the best indicators of GV. Methods：A total of 280 critically ill children admitted to the pediatric intensive care unit（PICU） of the Children’s Hospital of Chongqing Medical University from January 1st，2020 to August 31st，2020 were prospectively included in the study. The patients were divided into survivors or non-survivors according to the 28-day outcomes after PICU admission. Physiological parameters and outcomes data were recorded for 72 h after PICU admission，and the statistical analyses were made by several statistical methods such as the receiver operating characteristic（ROC） curve and logistic regression analysis. Results：①In different indicators of GV，glycemic standard deviation（SD）（AUC=0.759，95%CI=0.688-0.830，P<0.001） and glycemic lability index（GLI）（AUC=0.752，95%CI=0.680-0.825，P<0.001） were superior to the others for the prognosis. Multivariate logistic regression analysis also showed that SD（OR=3.356，95%CI=1.536-7.332，P=0.002），PIM（pedi－atric index of mortality） scores（OR=1.155，95%CI=1.023-1.303，P=0.020） and stay of PICU（OR=0.895，95%CI=0.837-0.956，P=0.001） showed an independent association with PICU 28-day mortality. ②Different from adult studies，there was a heteroge－neous effect of GV on children，and subgroup analysis disaggregated by tertiles of Mean and GV revealed that critically ill children in the lowest tertile of the Mean and in the highest tertile of GV had significantly highest mortality. Conclusion：GV can increase the risk of mortality in critically ill children，and among all glucose parameters，SD and GLI may be the dominant predictors representing for GV. Clinicians should pay more attention to GV as well as average blood glucose.