Abstract Adoptive Cell Transfer (ACT) of tumor infiltrating lymphocytes (TILs) is a promising strategy for cancer immunotherapy. The current therapy relies on non-myeloablative lymphodepletion and high dose IL-2 therapy, which has serious side effects for patients. Optimization of the current ACT therapy, using animal models, may help reduce or eliminate these additional treatments and enhance the response rate. ACT has been evaluated in the autoimmune model RIP-GP mice, which express the glycoprotein (GP) of lymphocytic choriomeningitis virus (LCMV) on the β-islet cells of the pancreas. Our findings showed that LCMV memory T cells stimulated with matured dendritic cells (DCs) and pulsed with peptides derived from LCMV, successfully induced diabetes. These stimulated T cells expanded well in vivo and showed strong killing activity after transfer. However, unstimulated memory tissue specific T cells rarely expanded after transfer and showed limited cytolytic activity. To evaluate this method in the tumor model, RIP-Tag2/GP double transgenic mice were used, which express the large T antigen under the control of RIP and spontaneously develop insulinomas. Islet specific T cells cocultured with DCs and LCMV-GP peptides displayed strong anti-tumor activity after transfer in vivo. Surprisingly, mice bearing insulinomas survived longer after ACT than mice treated with LCMV vaccination. Therefore, stimulating TILs before infusion has high potential for enhancing ACT therapy.