Summary Objectives The study investigated cross-sectional predictors of ever-undergone influenza vaccination (IV) and longitudinal predictors of first-time IV among Chinese elderly in Hong Kong. Methods A random telephone survey interviewed 886 Chinese respondents aged 65 and above and 483 of these 886 respondents (54.5%) completed another follow-up questionnaire. Results Of the 483 respondents, 25.1% (or 121) had ever undergone IV at baseline; 13% (47 of 362) were vaccinated for the first time during the follow-up period. The cross-sectional data identified 10 significant variables related to the Health Belief Model (HBM) predicting having ever undergone IV (e.g., perceived efficacy of prevention, side effects, financial difficulty, univariate OR = 1.58–68.14 and 0.31–0.47). None of these variables could prospectively predict first-time IV during the follow-up period; the only significant variable was whether the respondent visited social centers during the follow-up period (OR = 2.74). Conclusions The 10 studied variables (e.g., perceived efficacy, perceived side effects) were predictive of whether ever undergone IV in the cross-sectional survey. These variables were, however, unable to predict first-time IV in the longitudinal study. Therefore, programs modifying these cross-sectional factors (e.g. change perceptions on efficacy and safety) may not be effective in promoting first-time IV among the elderly. Longitudinal intervention studies are warranted.