ABSTRACT While the use of intensive insulin therapy has reduced maternal and fetal mortality associated with type 1 and type 2 diabetes in pregnancy, the risk of diabetes-related pregnancy morbidity remains high. Rates of pre- existing diabetes in pregnancy in the United States have increased 30% over the past 5 years, driven by a dramatic increase in the prevalence of type 2 diabetes in the obstetric population. The risk of pregnancy complications in diabetes is directly related to hyperglycemia; thus, guidelines recommend lower glycemic targets in pregnancy than outside pregnancy. However, these intensified pregnancy targets are extremely difficult to attain due to the risk of hypoglycemia, dynamic changes in insulin resistance across gestation, and the constant vigilance required for intensive insulin self-management during pregnancy. Unfortunately, pregnant individuals have been systematically excluded from trials of novel automated insulin delivery systems that have the potential to address the challenges of glycemic management and improve outcomes in diabetes- affected pregnancy. The iLet Bionic Pancreas is an FDA-approved automated insulin delivery system that improves glycemic control and reduces the burden of diabetes management in people with type 1 diabetes. The device is distinguished from other available systems because it is initiated with only the patient’s weight and eliminates the need for carbohydrate counting or user-initiated correction boluses. However, like all other available automated insulin delivery systems, the device was designed for a non-pregnant glycemic target range of 70- 180 mg/dl rather than the pregnancy glycemic target range of 63-140 mg/dl. We now propose to make feasible modifications to the iLet device to achieve pregnancy-specific glycemic targets in type 1 and type 2 diabetes. In Aim 1, we will conduct a 5-week crossover trial in non-pregnant individuals (N=20, 10 each with type 1 and type 2 diabetes) to optimize iLet settings for the achievement of pregnancy-specific glycemic targets. In Aim 2, we will compare optimized pregnancy-specific iLet settings to usual care in 36 pregnant individuals (18 each with type 1 and type 2 diabetes) an 8-week random-order crossover trial. These investigations will establish the first pregnancy-specific mode for a commercially available automated insulin delivery system and will, for the first time, extend the use of automated insulin delivery in pregnancy to individuals with type 2 diabetes. This work has the potential to transform the care of diabetes in pregnancy and substantially improve maternal and neonatal health outcomes for the almost 40,000 pregnancies affected by type 1 and type 2 diabetes yearly in the United States.