The immunogenicity of the trivalent split-virus influenza vaccine was investigated among 70 healthy adults (mean age: 48.5, range: 36-68). The vaccine antigens were: A/Yamagata/32/89 (H1N1); A/Beijing/352/89 (H3N2); and B/Bangkok/163/90. Regarding the entire sample, the vaccine induced a tenfold or more rise on the average in the hemagglutination inhibition (HAI) antibody to each antigen. The response rates (greater than or equal to a fourfold rise) were about 90% or more among those with a prevaccination titer < or = 1:64 (equivalent to < or = 1:16 on the Western scale; in Japan, the HAI titers are expressed by the final, and not the initial, dilution of the serum; from hereon our findings will be expressed using the Japanese scale), whereas they were 0-50% at > or = 1:128. Thus, the prevaccination titer was negatively associated with antibody induction. The achievement rates (postvaccination titer > or = 1:128) among those with a prevaccination titer < 1:16 remained at 48-68%. Regarding the analysis of variance, a significant effect on antibody induction was indicated for the prevaccination titer (P < or = 0.002), but not for age (P > or = 0.425). The interaction between the prevaccination titer and age was significant for A/Yamagata (P = 0.030), while it was also suggestive for A/Beijing (P = 0.054): as age increased, those with no preexisting antibody (< 1:16) showed greater titer rises, in contrast to the smaller rises among those with a titer > or = 1:16. Based on the attack survey conducted separately, the vaccine efficacy on influenza-like illnesses with fever < or = 37 degrees C and > or = 37.5 degrees C was calculated to be 16% (95% confidence interval: -66% to 57%) and 37% (-55% to 74%), respectively.