Background/Aims: The principle treatment strategy for glycemic management in most care settings is reactive; monitor A1c levels and then react when the A1c exceeds the recommended optimal care goal with treatment intensification. Our goal was to assess the potential to improve diabetes performance measures through preventive strategies directed at patients who are at A1c goal but at high risk for disease progression and A1c relapse. Methods: Patients not meeting optimal care goals were partitioned into one of three different A1c trajectories: (a) FLAT - those that are consistently above optimal A1c goal, (b) Negative slope - those patients that are on an improvement trajectory, and (c) Positive slope - those that have previously been meeting A1c goals but who have relapsed (often due to medical issues, co-morbidities, psychosocial stress, or behavioral or medication adherence, or disease progression). We quantified the proportion of patients with diabetes who contribute to the relapse vector by identifying patients with diabetes and A1c tests in the last two years (9/1/2012 - 8/31/2014) and quantifying the proportion of patients who relapsed in year 2, stratified by A1c range and pharmacologic treatment in year 1. Results: 29,321 patients were identified with at least 2 diabetes diagnoses in years 1 and 2, with median A1c 7.4%. 8889 (30%) had an A1c >8% in year 2. Of 6321 patients with A1c of 7-7.9% in year 1, 2332 (36.9%) relapsed to >8% in year 2. Relapse was higher (43.2%) for patients medicated with sulfonylurea or insulin. Only 689/10,202 (6.7%) patients with A1c 8% in year 2. Discussion: We estimate that the phenomenon of A1c relapse accounts for 1/3 of all adults identified as having uncontrolled glucose on quality measures. Pro-active care strategies in high-risk patients close to goal (A1c 7-7.9%) to help them sustain control could reduce the proportion of patients not meeting optimal A1c goals. More systematic use of patient reported self-monitored blood glucose data could further help to identify patients who are relapsing or progressing. Further research is needed to test these hypotheses.