Immunotherapy is becoming the standard of care for melanoma. However, resistance to therapy is a major problem. Previously, we showed that tumor-specific, cytotoxic CD4⁺ T cells from tyrosinase-related protein 1 transgenic mice could overcome secondary resistance to recurring melanoma when anti-programmed cell death 1 ligand (PD-L1) checkpoint blockade was combined with either anti-lymphocyte-activated gene 3 (LAG-3) Abs or depletion of tumor-specific regulatory T (Treg) cells. In this study, we show that PD-L1 expressed by the host, not B16 melanoma, plays a major role in the early stages of exhaustion or primary resistance. We observed durable regression of melanoma in tumor-bearing PD-L1^-/-RAG^-/- mice with transfer of naive tumor-specific CD4⁺ T cells. However, exhausted tumor-specific CD4⁺T cells, which included tumor-specific Treg cells, failed to maintain durable regression of tumors in PD-L1^-/-RAG^-/- mice unless tumor-specific Treg cells were eliminated, showing nonredundant pathways of resistance to immunotherapy were present. Translating these findings to a clinically relevant model of cancer immunotherapy, we unexpectedly showed that anti-PD-L1 checkpoint blockade mildly improved immunotherapy with tumor-specific CD4+ T cells and irradiation in wild-type mice. Instead, anti-LAG-3 checkpoint blockade, in combination with tumor-specific CD4⁺ T cells and irradiation, overcame primary resistance and treated established tumors resulting in fewer recurrences. Because LAG-3 negatively regulates effector T cell function and activates Treg cells, LAG-3 blockade may be more beneficial in overcoming primary resistance in combination immunotherapies using adoptive cellular therapy and irradiation than blockade of PD-L1.