BACKGROUND: Fingerstick blood glucose measurement has become widespread in both hospital and prehospital settings. OBJECTIVE: To determine the accuracy of fingerstick blood glucose measurement in patients with poor peripheral perfusion (shock). METHOD: Results obtained during three methods of glucose analysis (fingerstick blood glucose measurement; bedside and laboratory glucose analysis) were examined prospectively on 38 patients from inpatient medical and surgical critical care units or the emergency department of a large tertiary care referral center. RESULTS: The means of the three glucose measurements were significantly different. Univariate analysis of the mean laboratory glucose value versus the mean fingerstick glucose value was significantly different. The mean venipuncture glucose measured by the bedside glucose meter versus the mean venous laboratory glucose was not significantly different. CONCLUSION: These results suggest that fingerstick blood samples should not be used for bedside glucose analysis in patients who may have inadequate tissue perfusion.