Subjects (n = 312) received either the human diploid cell rabies vaccine (HDCV) or the purified Vero cell rabies vaccine (PVRV) according to either two-injection (days 0 and 28) or three-injection (days 0, 7, and 28) primary regimens. They received a booster injection at 1 year. Rabies antibody levels were measured after the primary series and the booster and then each year for the next 10 years. The results confirm the superior long-term immunogenicity of the three-injection over the two-injection protocol. HDCV and PVRV in three doses were equally immunogenic. A booster injection at 1 year provides long-term seroconversion (titer > or = 0.5 IU/mL). Antibody titers 2 weeks after the 1-year booster allowed prediction of long-term immunity. Good responders, with titers > or = 30 IU/mL, were protected for at least 10 years. An algorithm for differentiation between good responders and poor responders with respect to vaccine booster strategies is proposed.