Although influenza causes more hospitalizations and deaths among American children than any other vaccine-preventable disease, deriving accurate population-based estimates of disease impact is challenging. Using 2 independent surveillance systems, we performed a capture-recapture analysis to estimate influenza-associated hospitalizations in children in Davidson County, Tennessee, during the 2003-2004 influenza season. The New Vaccine Surveillance Network (NVSN) enrolled children hospitalized with respiratory symptoms or fever and tested them for influenza. The Tennessee Emerging Infections Program (EIP) identified inpatients with positive influenza diagnostic test results through review of laboratory and infection control logs. The hospitalization rate estimated from the capture-recapture analysis in children <5 years of age was 2.4 per 1,000 (95% confidence interval 1.8-3.8). When NVSN estimates were compared with capture-recapture estimates, NVSN found 84% of community-acquired cases, EIP found 64% of cases in which an influenza rapid test was performed, and the overall sensitivity of NVSN and EIP for influenza hospitalizations was 73% and 38%, respectively.