Nationally representative estimates of diabetes prevalence published by the Centers for Disease Control and Prevention (CDC) are derived from self-reported cross-sectional surveys. Electronic medical records (EMRs) may allow for a more complete accounting of diabetes cases. The SUPREME-DM project combined EMR data from 11 HMO's to estimate annual diabetes prevalence in 2005-2009. In each year 2005-2009, we identified HMO members aged > 20 years who had at least 6 months of health plan membership. Using EMR data from 2000-2009, we identifi ed individuals who had 2+ laboratory test results indicative of diabetes, 1 inpatient or 2+ outpatient diabetes diagnoses, or any anti-diabetic medication dispensed other than during pregnancy. We estimated total prevalence and by age categories and sex, and calculated the change in prevalence between 2005 and 2009. From 2005 to 2009, the mean annual adult membership of the 11 HMO's was 7,029,226. Total diabetes prevalence increased by 17.4%, from 9.3% in 2005 to 10.9% in 2009. The highest prevalence in all years was seen among 75-84 year olds (25.1% to 29.3%). All age groups experienced large prevalence increases, ranging from 9.4% (aged 65-84) to 25.5% (aged 85+). Prevalence was higher among men than women in all years (e.g., 12.0% vs. 10.9% in 2009).The diabetes prevalence rates we report are higher than previous CDC estimates (e.g., 8.3% vs. 10.9% overall in 2009). This may be due in part to more comprehensive case ascertainment using EMRs. Our results suggest that the burden of diabetes in the U.S. may be greater than previously believed and continues to grow steadily.