Checkpoint inhibitors such as PD-1/PD-L1 antibody therapeutics are a promising option for the treatment of multiple cancers. Due to the inherent limitations of antibodies, great efforts have been devoted to developing small-molecule PD-1/PD-L1 signaling pathway inhibitors. In this study we established a high-throughput AlphaLISA assay to discover small molecules with new skeletons that could block PD-1/PD-L1 interaction. We screened a small-molecule library of 4169 compounds including natural products, FDA approved drugs and other synthetic compounds. Among the 8 potential hits, we found that cisplatin, a first-line chemotherapeutic drug, reduced AlphaLISA signal with an EC₅₀ of 8.3 ± 2.2 μM. Furthermore, we showed that cisplatin-DMSO adduct, but not semplice cisplatin, inhibited PD-1/PD-L1 interaction. Thus, we assessed several commercial platinum (II) compounds, and found that bis(benzonitrile) dichloroplatinum (II) disturbed PD-1/PD-L1 interaction (EC₅₀ = 13.2 ± 3.5 μM). Its inhibitory activity on PD-1/PD-L1 interaction was confirmed in co-immunoprecipitation and PD-1/PD-L1 signaling pathway blockade bioassays. Surface plasmon resonance assay revealed that bis(benzonitrile) dichloroplatinum (II) bound to PD-1 (K_D = 2.08 μM) but not PD-L1. In immune-competent wild-type mice but not in immunodeficient nude mice, bis(benzonitrile) dichloroplatinum (II) (7.5 mg/kg, i.p., every 3 days) significantly suppressed the growth of MC38 colorectal cancer xenografts with increasing tumor-infiltrating T cells. These data highlight that platinum compounds are potential immune checkpoint inhibitors for the treatment of cancers.