Résumé : Objective: To determine the safety and efficacy of a change in blood glucose (BG) control protocol from a single target to 2 targets based on diabetes mellitus (DM) status and glycated hemoglobin A1C (A1C) in a cohort of critically ill patients. Methods: This investigation includes 1,979 patients admitted to a single intensive care unit (ICU) between September 16, 2013 and September 15, 2015. The BG target was 90 to 120 mg/dL in the PRE era and 80 to 140 mg/dL for patients without diabetes (NON) and with DM with A1C <7% and 110 to 160 mg/dL for DM with A1C ≥7% (TIGHT and LOOSE protocols) in the POST era. The primary efficacy outcome was the observed:expected (O:E) mortality ratio. Results: Among NON, the mean BG was slightly lower in the POST era: 118 (106-132) versus 115 (101-120) mg/dL (P =.0003). Among DM, the mean BG was 139 (123-160) mg/dL in the PRE era versus 136 (119-149) and 159 (138-171) mg/dL for TIGHT and LOOSE in the POST era (P =.0668 and.0001, respectively). Overall, 11.0% and 11.8% of patients had at least 1 BG level <70 mg/dL in the 2 eras (P =.68). The O:E mortality ratios for NON and DM for the PRE and POST eras were 0.75 versus 0.74 (P =.51) and 0.69 versus 0.52 (P<.001) respectively, and among DM with A1C ≥7% were 0.74 versus 0.52 (P =.004). Conclusion: This hypothesis-generating investigation suggests the need for additional prospective interventional studies assessing the outcomes of patients randomized to personalized glucose targets.