Aims To assess the differences of long-term efficacy between plasma-derived and recombinant hepatitis B virus (HBV) vaccines and the effectiveness of catch-up vaccination in adolescents with undetectable anti-HBs. Methods Before 1992, infants born in Taiwan were immunized using plasma-derived HB vaccine, and thereafter, by using recombinant HB vaccine. From the only junior middle school of a rural township in central–southern Taiwan, 1788 (93.7%) students from five cross-sectional screenings, grouping into three birth cohorts (Group I: born during 1984–1986, II: 1986–1992 and III: 1992–1995), were enrolled for checking HBsAg, anti-HBs and anti-HBc. Students with undetectable HBsAg and anti-HBs underwent a booster dose (2.5 ug) of recombinant HB vaccine (Engerix-B; GlaxoSmithKline, Rixensart, Belgium) and had anti-HBs re-checked 3 weeks later. Individuals who had remained undetectable for anti-HBs completed the other two doses of HB vaccines at 1 and 6 months later. Results The prevalence of HBsAg (11.4, 5.4 and 1.2%), anti-HBs (64.5, 44.1 and 36.0%) and anti-HBc (29.5, 12.5 and 4.4%) decreased from Group I to III ( P < 0.001 for trends). After a booster dose, the positive rates of anti-HBs increased up to 80.5% (16% increase) in Group I, 81.0% (36.9% increase) in Group II, and 94.4% (58.4% increase) in Group III. The percentages of anamnestic response increased with a trend ( P < 0.001). A total of 110 non-responders completed 3 doses of catch-up HB vaccination, but 3 cases (2.7%) of Group II, evoked primary vaccination response. Conclusion Recombinant vaccine showed predominant disappearance rate (62.7%) of anti-HBs 12–15 years after vaccination, but provided better anamnestic response after a booster dose. It also showed high success rate (97.3%) in catch-up vaccination in adolescents.