A meta-analysis of randomized and quasi-randomized control trials evaluating influenza vaccination efficacy in healthy adults was carried out to investigate potential sources of heterogeneity among the published studies. The pooled estimates of vaccine efficacy were 22% (95% CI = 16-28) for preventing clinically diagnosed cases of influenza and 63% (95% CI = 53-71) for preventing laboratory confirmed cases of influenza. These estimates are likely to vary substantially according to the age of the subjects, with vaccine efficacy decreasing significantly with the increase of the mean age of the patients, and to the methodological quality of trials, with low quality trials associated with an overestimation of influenza vaccine efficacy. Some evidence of publication bias was found, particularly when the outcome assessed was clinically diagnosed influenza. Further high-quality clinical trials, designed to facilitate future pooled analyses, are needed to provide definitive answers for policy-makers.