Adoptive transfer of T cells that have been genetically modified to express an antitumor T-cell receptor (TCR) is a potent immunotherapy, but only if TCR avidity is sufficiently high. Endogenous TCRs specific to shared (self) tumor-associated antigens (TAAs) have low affinity due to central tolerance. Therefore, for effective therapy, anti-TAA TCRs with higher and optimal avidity must be generated. Here, we describe a new in vitro system for directed evolution of TCR avidity using somatic hypermutation (SHM), a mechanism used in nature by B cells for antibody optimization. We identified 44 point mutations to the Pmel-1 TCR, specific for the H-2Db -gp10025-33 melanoma antigen. Primary T cells transduced with TCRs containing two or three of these mutations had enhanced activity in vitro. Furthermore, the triple-mutant TCR improved in vivo therapy of tumor-bearing mice, which exhibited improved survival, smaller tumors and delayed or no relapse. TCR avidity maturation by SHM may be an effective strategy to improve cancer immunotherapy.