The objective was to assess the effect of online use of a patient portal on improvement of glycohemoglobin (HbA1c) in patients with type 2 diabetes presenting to primary care clinics. This retrospective cohort design used data from a primary care patient data registry that captured all ambulatory visits to the academic medical center's primary care clinics. A total of 1510 patients with diabetes were included because they had at least 1 visit with a documented HbA1c value between January 1, 2010, and June 30, 2013. Degree of patient portal use was defined as no use, read only, and read and write. Linear regression models were computed to measure the association between degree of patient portal use and HbA1c control before and after adjusting for demographics, comorbidity, and volume of health care use. Patients who were nonusers of the patient portal's e-mail function had consistently higher average HbA1c values than patients who read and wrote e-mails. After adjusting for demographics, health services utilization, and comorbid conditions, patients who read and wrote e-mails still had significantly (P<0.001) lower average HbA1c values compared to nonusers (ß=-0.455; 95% confidence interval [CI]:-.632-.277). In adjusted analysis, patients who only read e-mail also had significantly (P<0.05) lower mean HbA1c values compared to nonusers (ß=-0.311, 95%CI:-.61--0.012). Patients with more active e-mail communication via a patient portal appeared to have the greatest likelihood of HbA1c control. Patients should be encouraged to use this resource as a means of communication with providers and not merely a passive source of information. (Population Health Management 2016;19:125-131).