The accuracy of glucose test strip in critically care has been questioned. We investigated the accuracy of glucose test strip in critically ill children. Patients, aged from 1 month to 18 years admitted in pediatric intensive care unit. Demographic data, hemodynamic parameters, and perfusion index (PI) were recorded. Glucose test strips were performed from finger stick blood [capillary blood glucose (CBG)] and from whole blood [whole blood glucose (WBG)] along with laboratory plasma blood glucose (PBG) from either arterial or venous blood samples. The accuracy of glucose test strips was defined according to ISO 15197 and Clarke error grid (CEG). One hundred and eighty one blood samplings including 117 arterial blood (CBG, WBGa, PBGa) and 64 venous blood (CBG, WBGv, PBGv) were obtained. The accuracy of WBGa was 98.3 and 95.2% when compared to the accuracy of CBG (88.7 and 83.3%. The accuracy of WBGv was 92.2% and 87.0 when compared to the accuracy of CBG which was 79.7 and 72.9% (ISO 15197: 2003 and 2013, respectively). Bland-Altman plot demonstrated bias and precision of 7.4±17.7 mg/dL in acceptable PI group compared to 30.2±23.4 mg/dL in low PI group (PI≤0.3). The CBG test strip must be interpreted carefully in critically ill children. A low PI was associated with poor CBG strip accuracy. WBG test strip from arterial blood was more appropriate for glucose monitoring in children with peripheral hypoperfusion.