Anti-PD-1/PD-L1 immune checkpoint blockade (ICB) therapy has revolutionized the treatment of many types of cancer over the past decade. The initial therapeutic hypothesis underlying the mechanism of anti-PD-1/PD-L1 ICB was built around the premise that it acts locally in the tumor, reversing the exhaustion of PD-1CD8 T cells by "releasing the brakes." However, recent studies have provided unprecedented insight into the complexity within the CD8 T-cell pool in the tumor microenvironment (TME). Single-cell RNA sequencing and epigenetic profiling studies have identified novel cell surface markers, revealing heterogeneity within CD8 T-cell states classified as unique. Moreover, these studies highlighted that following ICB, CD8 T-cell states within and outside the TME possess a differential capacity to respond, mobilize to the TME, and seed an effective antitumor immune response. In aggregate, these recent developments have led to a reevaluation of our understanding of both the underlying mechanisms and the sites of action of ICB therapy. Here, we discuss the evidence for the reversibility of CD8 T-cell exhaustion after ICB treatment and its implication for the further development of cancer immunotherapy.