Because severity of disease, duration, and co-morbidities varies in patients with type 2 diabetes, a single treatment strategy is not effective for all patients. In this study, we investigate how to best match various treatment approaches to different types of patients, and to assess the impact of personalized care strategies on health outcomes and cost. We examine two treatment strategies inspired from control theory: feedback (FB) and feedforward (FF). Strategy FB makes treatment decisions based on the current patient state. Strategy FF bases decisions on the anticipated future state. We compare these two strategies with use of an existing treatment guideline, i.e., Staged Diabetes Management® (SDM). A major difference between the strategies is visit frequency (maximum # of visits for SDM=12, FF=6, FB=4). We used a previously developed care simulation model to apply the 3 strategies for 1 year to a population of 10000 simulated patients modeled after a real patient population with type 2 diabetes. Cost effectiveness of strategies was assessed by calculating the average amount (out-patient) spent on treatment of a group to bring a patient to goal (Total Cost / Patients at Goal, CPG). As shown in Table 1, patients are grouped according to their initial A1c into: low (<8%), medium (8-10%), and high (>10%) A1c categories. SDM is the most cost-effective strategy for the patients in the high A1c group. Strategies FB and FF are the most effective for the low and medium A1c groups respectively, leading to total savings of $2.1M in treatment costs in 1 year. We conclude that costs and clinical quality of care vary using different personalized treatment approaches. The best method (SDM, FF, or FB) varies depending on the clinical situation and baseline A1c level.