Among 2099 uninfected subjects in phase I and II trials of candidate AIDS vaccines, 23 were diagnosed with intercurrent human immunodeficiency virus type 1 (HIV-1) infection. High-risk sexual exposures accounted for 17 infections, and intravenous drug use accounted for 6. Four subjects received placebo, 13 received a complete immunization schedule (> or = 3 injections), and 6 were partially immunized (< or = 2 injections). There was no significant difference between vaccine recipients and control groups in incidence of HIV-1 infection, virus load, CD4 lymphocyte count, or V3 loop amino acid sequence. In summary, 19 vaccinated subjects acquired HIV-1 infection during phase I and II trials, indicating that immunization with the products described is < 100% effective in preventing or rapidly clearing infection. Laboratory analysis suggested that vaccine-induced immune responses did not significantly affect the genotypic or phenotypic characteristics of transmitted virus or the early clinical course of HIV-1 infection.