LT-85 is an alveologenic adenocarcinoma of C3Hf/HeN mice. Comparisons of the in vitro and in vivo surface properties of these cells revealed that under normal conditions, they expressed I-A and I-E antigens iv vivo only. By using clonally derived cells, it was established that this phenomenon was not due to the selection of an Ia antigen-positive tumor cell subpopulation, but resulted from phenotypic conversion of Ia antigen-negative tumor cells. These tumor cells and 1053 cells (a fibrosarcoma of C3H/HeN MTV- mice) could, however, be induced to express I-A, I-E, and much higher levels of H-2 antigens in vitro by co-culturing them with spleen cells from LT-85 tumor-bearing C3H/HeN MTV- mice. In vitro induction of Ia and H-2 antigens did not result from contaminating splenocytes or from antigen transfer, because splenocytes from BALB/c (H-2d) mice immunized with A/J (H-2k/d) cells were able to induce the expression of Iak antigens by both tumor cell lines. It was found that this phenomenon was neither H-2-restricted nor antigen-specific. The results clearly indicated, however, that an immune response was required to generate phenotypic conversion of the tumor cells, both in vivo and in vitro. It was further found that soluble, rather than cellular, factors produced during an immune response induced the expression of Ia antigens by LT-85 and 1053 tumor cells. In contrast to what has been reported about the induction of Ia antigens on macrophages and normal epithelial and endothelial cells, the induction of Ia antigens on LT-85 and 1053 cells did not appear to require T cells, and did not involve gamma-interferon. These findings demonstrate that some tumor cells are capable of altering their MHC antigen phenotype in response to factors produced during an immune response in vivo or in vitro. Because of the involvement of Ia antigens in several aspects of immune phenomena, the ability of tumor cells to differentially express Ia antigens in response to environmental factors may have profound effects on host-tumor interactions. Furthermore, the differences seen in the phenotypes of tumor cells grown in vitro and in vivo suggest that in vitro methodologies of tumor cell characterization may not present a complete picture of the natural state of the tumor cell surface.