Models predict that immunizing as few as 20% of school children, important transmitters of influenza, will reduce influenza-related illness in the elderly. We evaluated the potential herd immunity during three influenza seasons, 2005-2006, 2006-2007 and 2007-2008, which followed the immunization of>40% of school children in Knox County (KC), TN, with live, attenuated influenza vaccine. Individual-level demographic, health status and health service utilization information about KC residents>65 years and those residing in the 8 surrounding counties was obtained from the United States Medicare Program's administrative data. Influenza seasons were identified based on virus isolation. Pneumonia and influenza (P&I) hospitalization rates per 1,000 were compared between the elderly residing in the two areas for the three influenza seasons, and the 3 prior seasons. Differences-in-difference multivariate analysis allowed us to estimate the effect of the school-based immunization program on P&I hospitalization rates simultaneously adjusting for other important individual-level covariates. The age-adjusted rates among the KC residents were significantly lower, 4.62 and 6.02 versus 6.54 and 7.58 than in the residents of the comparison counties during the first two intervention seasons, p=0.001 and 0.037, respectively, but not in the third. However, after adjusting for the traditionally lower rates of P&I hospitalization in the comparison counties, as well as for the other covariates, we were not able to demonstrate a statistically significant effect of the vaccination program in reducing the rates in either group of the elderly. The impact of the covariates was as expected. Those associated with increased P&I hospitalization rates were increasing age, lower income, poorer health status, prior hospitalization (particularly for P&I), and high prior use of physician services. Influenza immunization of an elderly person reduced his/her probability of being hospitalized for P&I. In conclusion, Immunization of>40% of school children did not result in a reduction of P&I hospitalization rates among the elderly. We believe that the failure to show an impact was likely due to the high level of immunization among the elderly (> 60%). Administration of influenza vaccine to children as a way to protect the elderly in situations where vaccine supplies are limited or the elderly are not accustomed to receiving influenza vaccine may still be appropriate.