T lymphocytes recognize a limited number of antigenic sites on any given antigenic protein. We find that the presence or absence of a response to one immunodominant site can make the difference between a high responder and a low responder, even though low responders respond to other sites almost as well as high responders. Besides interaction with major histocompatibility (MHC) molecules, we find that the mode by which the antigen is processed into fragments for T-cell recognition also determines which sites are seen. The products of natural processing of the protein appear to be larger than the synthetic peptides and contain structures which hinder binding to certain MHC molecules or to the T-cell receptor. Besides MHC binding and antigen processing, a third factor in immunodominance is the intrinsic structure of the antigenic site. We have previously shown that amphipathic helices have a high chance of being immunodominant, and have developed a computer program to locate such structures in protein sequences. We prospectively predicted sites in the malaria circumsporozoite protein and found that the four most widely recognized sites in an endemic area of West Africa were all predicted. Similarly, we identified two helper T-cell sites from the HIV (AlDS virus) envelope, and showed that immunization with these elicits enhanced antibody responses to the whole envelope when injected into monkeys. These sites are also recognized by human T cells from volunteers who have been immunized with a recombinant vaccinia virus expressing the HlV envelope. In mice, we have now identified other immunodominant sites in the envelope protein, including a very large one within which different MHC types of mice each recognize different overlapping regions. Also, since cytotoxic T lymphocytes (CTL) may play a critical role in defense against AlDS, we have used a recombinant vaccinia virus and transfectants expressing the HIV envelope gene to induce specific CTL against the HIV envelope. Using synthetic peptides, we were able to identify the first known CTL recognition site in the AIDS virus. However, we observed that such sites are very limited. These results may hopefully contribute_to the design of a vaccine for AIDS.