We have examined the frequency and alloantigen specificity of CTL that accumulate in sponge allografts (sponges seeded with allogeneic splenocytes) in sponge isografts (sponges seeded with syngeneic splenocytes), and in splenocyte-free sponge implants. Using limiting dilution analysis (LDA), we observed that sponge isografts and splenocyte-free sponge implants from C57BL/6 (H-2b) mice usually acquire small numbers of CTL (less than 250 cells per graft) with DBA/2 (H-2d)-reactivity or C3H/HeJ (H-2k)-reactivity. These alloreactive CTL are not detectable in conventional 51Cr-release assays, presumably because they are too infrequent and/or because they are inactive CTL precursors. When we examined the accumulation of alloreactive CTL in sponge allografts, we observed that DBA/2 sponge allografts from C57BL/6 recipients accumulate 10 to 100 times more DBA/2-reactive CTL than alloantigen-free sponge grafts. Nonetheless, these donor-reactive CTL rarely constitute more than 0.5% of the T cells recovered from sponge allografts, even at the peak of the rejection response. This raises questions concerning the remaining 99.5% of the allograft-infiltrating T cells. We were unable to detect by LDA any host-reactive CTL in sponge allografts, thus excluding the possibility that some of the remaining T cells were host-reactive CTL of donor origin which diluted graft-reactive T cells. However, using LDA we did detect a significant number of third-party (C3H/HeJ)-reactive CTL in sponge allografts, suggesting that the intense immune response at a graft site might facilitate indiscriminate recruitment of T lymphocytes. Alternatively, this enhanced third-party alloreactivity might reflect the proliferation of donor-reactive CTL with incidental crossreactivity for C3H/HeJ alloantigens. While testing these two alternatives, we observed that LDA cultures designed to detect third-party-reactive CTL could also support the growth of the in vivo-activated, donor-reactive CTL from sponge allografts; This compromised enumeration by LDA of the less frequent, third-party-reactive CTL by LDA. Although LDA is the only method that detects the growing population of third-party-reactive CTL in sponge allografts, technical restraints exclude LDA as a method of determining whether donor-reactive CTL and third-party-reactive CTL are separate or overlapping CTL subpopulations. Hence, it remains unclear if third-party-reactive CTL are a significant or insignificant proportion of the CTL that infiltrate sponge allografts.(ABSTRACT TRUNCATED AT 400 WORDS)