OBJECTIVES: To determine which factors were associated with sustained poor or good glycemic control over three years. METHODS: This retrospective study of enrollees in a large health plan in Hawaii with diabetes was restricted to those with poor control at baseline (HbA1c>9% in the year 2006), who had HbA1c values for four years (2006-2009, n=2970).  Separate multivariable logistic regressions examined factors related to poor (HbA1c>9%) or good (HbA1c<7%) control over three years.  Independent variables included age, gender, island of residence, type of coverage, comorbid conditions, diabetes duration, and number of medications. RESULTS: Relative to patients with diabetes duration of a year or less, patients with diabetes for >8 years were significantly less likely to sustain good control [OR=0.13, 95%CI(0.09, 0.21)], as were patients with diabetes for 6-7 years [OR=0.23, 95%CI(0.13, 0.39)], and 4-5 years [OR=0.17, 95%CI(0.09, 0.31)]. Compared to patients in HMO, patients in PPO or Medicare cost contract were significantly more likely to have good control.    Factors significantly related to poor control were age, diabetes duration, morbidity level, and number of medications.  Relative to patients aged 50-64, younger patients were significantly more likely to have poor control [age<35 OR=1.98, 95%CI(1.25, 3.14); age 35-49 OR= 1.27, 95%CI(1.03, 1.57)] and older patients were significantly less likely [age 65-79 OR=0.75, 95%CI(0.57, 0.98); age 80+ OR=0.63, 95%CI (0.41, 0.97)].  Longer diabetes duration and higher morbidity increased the risk of poor control, while patients taking 15+ medications decreased the risk. CONCLUSIONS: As diabetes duration strongly predicted poor control, primary prevention to decrease early onset of diabetes might be a cost-effective means to reduce the number of patients with poor control.  Our findings that younger patients with diabetes are most likely to have poor control suggest that interventions need to be targeted to this population.