BACKGROUND: The 2009 novel H1N1 influenza epidemic generated interest in regional and national influenza surveillance methods. Some systems revamped traditional syndromic and laboratory surveillance techniques, whereas others tracked influenza by Internet-based searches or other unique methods. We hypothesized that an influenza-like illness (ILI) surveillance system at a single university hospital would be accurate and useful for monitoring local influenza activity and impact. METHODS: We developed a system of ILI surveillance at 8 sentinel sites associated with a university health care system before the pandemic 2009-10 influenza season. Most sentinel sites used a symptom-based definition of ILI, whereas others used electronic medical records-based definitions. RESULTS: Results of the local ILI surveillance network correlated well with the onset and peak of the influenza season compared with state and regional ILI data, closely approximated cases of microbiologically confirmed influenza, demonstrated early onset of illness in one site in the sentinel site network, and were available several days sooner than data from existing surveillance systems. CONCLUSIONS: Local influenza surveillance at a single-institution level provided timely, useful, and accurate information, which helped guide resource utilization during the pandemic influenza season. The system was an important supplement to state and regional influenza surveillance.