The efficacy and safety of a recombinant yeast-derived hepatitis B vaccine were evaluated in 209 subjects after three administrations at 0, 4 and 20 weeks. Subjects were divided into four groups given 5 micrograms vaccine subcutaneously, 10 micrograms subcutaneously, 10 micrograms intramuscularly and 20 micrograms subcutaneously to define the effective dose and to compare the effect of administration. Seroconversion of the antibody to hepatitis B surface antigen after the third vaccination reached 96.6% in the group given 5 micrograms vaccine subcutaneously and 100% in the other groups. The final geometric mean antibody titres were 700 IU/l in subjects given 5 micrograms subcutaneously, 2004 IU/l in those given 10 micrograms subcutaneously, 4674 IU/l in those given 10 micrograms intramuscularly and 3342 IU/l in those given 20 micrograms subcutaneously. In the groups given 10 micrograms, the early seroconversion rate of the antibody to hepatitis B surface antigen and the geometric mean antibody titres after the third vaccination were significantly higher in subjects administered intramuscularly than subcutaneously (P less than 0.05). No major adverse effects were observed and minor reactions were the same as, or less than, those reported for the plasma-derived vaccine. Before and after administration, no significant fluctuation in the yeast antibody titre was observed. These results demonstrate the efficacy and safety of the yeast-derived vaccine, and show that 10 micrograms was the effective dose.