The fundamental issue in assessing influenza vaccine efficacy is to observe all study subjects with equal intensity throughout the surveillance period. The case definition can be adopted within the scope of the budget and the logistics of the study; however, there is no doubt that culture-proven influenza is currently the best outcome index. More pronounced vaccine efficacy can be detected if stricter case definition criteria are applied and/or if observations are confined to the peak epidemic period. Patients identified through passive case-finding in clinics do not properly represent all influenza cases that occur in the study subjects. Almost all non-randomized studies which have so far been conducted by Japanese clinicians do not take confounders into consideration. Even though laboratory-confirmed influenza is identified, vaccine efficacy should primarily be estimated based on the presence of any influenzal illness, since efficacy calculated by virus type or subtype often results in loss of statistical power. The results from post hoc subgroup analysis may not offer a solid base for assessing vaccine efficacy and should be cautiously interpreted.