Targeting immune checkpoint receptors on T cells is a common cancer treatment strategy. Frequently, this is accomplished through antibodies targeting the ligand of inhibitory co-receptors. Blocking the immune checkpoint PD-1 binding to its ligands PD-L1 and PD-L2 prevents downstream signaling and enhances anti-tumor T cell responses. This approach improved cancer patients' outcome. However, only one-third of the patients respond to these treatments. To better understand the mechanism of anti-PD-1 antibodies, we explored the location of PD-1 within the immune synapse. Surprisingly, we discovered that anti-PD-1 antibodies, besides blocking the interaction between PD-1 and its ligands, also removed PD-1 from the synapse. We demonstrated a correlation between removing PD-1 from the synapse by anti-PD-1 antibodies and the extent of T cell activation. Interestingly, a short version of the anti-PD-1 antibody, F(ab') 2 , failed to remove PD-1 from the synapse and activate T cells. Using syngeneic tumor model, we showed a superior anti-tumor effect to anti-PD-1 antibody over the shorter version of the antibody. Our data indicates that anti-PD-1 antibodies activate T cells by removing PD-1 away from the synapse and changing the location of PD-1 or other immune receptors within immune synapse could serve as an alternative, efficient approach to treat cancer.