Adoptive T-cell therapy is clinically efficacious in the treatment of select cancers. However, it is often difficult to obtain adequate numbers of tumor-specific T cells for therapy. One method for overcoming this limitation is to generate tumor-specific T cells by retrovirally mediated T-cell-receptor (TCR) gene transfer. However, despite instances of therapeutic success, major obstacles remain, including attaining the survival of retrovirally modified T cells in vivo as well as inducing long-term and multi-gene retroviral expression. Using a murine model of adoptively transferred retrovirally modified CD8(+) T cells, where antitumor immunity was dependent on sustained, multigene expression, we found that in vitro assays are poor indicators of in vivo efficacy. Despite persisting for over 9 months in a nonlymphopenic environment, genetically modified T cells exhibited discordant retrovirally mediated gene expression in vivo not readily evident from initial in vitro assays. In particular, one of the two TCR subunit genes necessary for antigen specificity was selectively lost in vivo. As this discordant gene expression was associated with the loss of antitumor immunity, consideration of these findings may provide guidance in the design, evaluation and application of retroviral vectors for use in the treatment of cancer and other human disease.