Résumé : We compared the efficacy of immunization with either simian immunodeficiency virus (SIV) Env glycoprotein (Env), Env plus Gag proteins (Gag-Env), or whole inactivated virus (WIV), with or without recombinant live vaccinia vector (VV) priming, in protecting 23 rhesus macaques (six vaccine and two control groups) from challenge with SIVmac251 clone BK28. Vaccination elicited high titers of syncytium-inhibiting and anti-Env (gp120/gp160) antibodies in all vaccinated macaques and anti-Gag (p27) antibodies in groups immunized with WIV or Gag-Env. Only WIV-immunized macaques developed anticell (HuT78) antibodies. After homologous low-dose intravenous virus challenge, we used frequency of virus isolation, provirus burden, and change in antibody titers to define four levels of resistance to SIV infection as follows. (i) No infection ('sterilizing' immunity) was induced only in WIV-immunized animals. (ii) Abortive infection (strong immunity) was defined when virus or provirus were detected early in the postchallenge period but not thereafter and no evidence of virus or provirus was detected in terminal tissues. This response was observed in two animals (one VV-Env and one Gag-Env). (iii) Suppression of infection (incomplete or partial immunity) described a gradient of virus suppression manifested by termination of viremia, declining postchallenge antibody titers, and low levels (composite mean = 9.1 copies per 106 cells) of provirus detectable in peripheral blood mononuclear cells or lymphoid tissues at termination (40 weeks postchallenge). This response occurred in the majority (8 of 12) of subunit-vaccinated animals. (iv) Active infection (no immunity) was characterized by persistent virus isolation from blood mononuclear cells, increasing viral antibody titers postchallenge, and high levels (composite mean = 198 copies per 106 cells) of provirus in terminal tissues and blood. Active infection developed in all controls and two of three VV-Gag-Env-immunized animals. The results of this study restate the protective effect of inactivated whole virus vaccines produced in heterologous cells but more importantly demonstrate that a gradient of suppression of challenge virus growth, reflecting partial resistance to SIV infection, is induced by subunit vaccination. The latter finding may be pertinent to studies with human immunodeficiency virus vaccines, in which it is plausible that vaccination may elicit significant suppression of virus infection and pathogenicity rather than sterilizing immunity.