OBJECTIVES: In the context of an epidemic, public health authorities need to have real time information of the disease propagation to make appropriate and timely decisions. We present our experience in using centralized electronic patient records in the early detection and follow-up of Flu A/H1N1 epidemic in France. METHODS: We used Longitudinal Patient Data (LPD), which is a dynamic centralized database of more than 1.3 million anonymous patient records uploaded regularly by a network of 1300 representative general practitioners and pediatricians in France. The trends of seasonal influenza, Flu A/H1N1 and flu-like syndrome diagnoses are traced and compared with the reports published by Sentinel Network, a group of trained physicians set up by public authorities. RESULTS: The trends of seasonal influenza, Flu A/H1N1, and flu-like syndrome obtained from LPD database show a very close similarity with those published by Sentinel Network (Pearson’s correlation coefficient = 0.97). The seasonal pattern of the incidence of influenza and flu-like syndrome were less similar between 2009 and 2008 (Pearson’s correlation coefficient = 0.60), which can be explained by an over declaration of all kinds of flu by the doctors during the second half of 2009. There was also a close similarity between seasonal Flu, approved Flu A/H1N1, and flu-like syndrome trends with a peak incidence in late November 2009. CONCLUSIONS: The LPD data matched very closely the results published by the official Sentinel Network. The advantage of the LPD data is that its trends are available almost in real time, whereas the active reporting of the doctors of Sentinel Network is done on a weekly basis. However, the LPD trends lack of geographical precision because of the censorship of the post codes of patient during the de-identification process. We recommend a combined use of LPD and Sentinel Networks in the follow up of future epidemics.