Mixed leukocyte cultures (MLC) between NZB responder spleen cells and Qa-1 disparate stimulator spleen cells were employed to determine the cellular requirements for the generation of primary anti-Qa-1 cell-mediated lympholysis (CML) responses. Although primary anti-Qa-1 cytotoxic lymphocytes (CTL) were generated during H-2-homologous stimulation, anti-Qa-1 CTL were not detectable from MLC in which the stimulators were H-2 allogeneic. Anti-Qa-1 CTL also were not generated from MLC in which the stimulators were semiallogeneic. Thus, H-2 identity between responder and stimulator cells was not sufficient to permit the generation of primary anti-Qa-1 CTL when H-2 disparity was also present. The capacity for H-2 disparity to prevent anti-Qa-1 CML responses was further demonstrated in MLC containing both H-2-allogeneic and H-2-homologous stimulator cells. Therefore, in subsequent studies we employed NZB responders and H-2-homologous, Qa-1-disparate stimulators. When various subpopulations of stimulator cells were studied for their ability to induce anti-Qa-1 CTL, nylon wool-adherent cells were found to be potent stimulators, but nylon wool-nonadherent cells were not. Furthermore, depletion of macrophages from the stimulator population abrogated the generation of anti-α-1 CML responses, despite the presence of responder macrophages in the culture. In contrast, all fractionated subpopulations stimulated anti-H-2 CML responses. When macrophage-enriched cells were used as stimulators, anti-α-1 CTL could be generated with approximately 80-fold fewer stimulator cells than when unfractionated splenocytes were used as stimulators. These findings indicated that stimulator macrophages were essential for the generation of primary anti-α-1 CTL. Direct evidence for macrophage expression of Qa-1-antigens was obtained by using a Qa-1(b)-specific CTL clone. These studies provide i) the first evidence for Qa-1 expression on macrophages, ii) a basis for comparison of the cellular interactions necessary to generate CTL against H-2K/D-encoded vs Qa-1-encoded class 1 antigens, and iii) a model for investigating the mechanisms responsible for the immunodominance of H-2K/D alloantigens.