Abstract Shelterin, a protective complex at telomeres, plays essential roles in cancer. In addition to maintain telomere integrity, shelterin functions in various survival pathways. However, the detailed mechanisms of shelterin regulation in cancer remain elusive. Here, we perform a comprehensive analysis of shelterin in 9125 tumor samples across 33 cancer types using multi-omic data from The Cancer Genome Atlas, and validate some findings in Chinese Glioma Genome Atlas and cancer cell lines from Cancer Cell Line Encyclopedia. In the genomic landscape, we identify the amplification of TRF1 and POT1, co-amplification/deletion of TRF2–RAP1–TPP1 as the dominant alteration events. Clustering analysis based on shelterin expression reveals three cancer clusters with different degree of genome instability. To measure overall shelterin activity in cancer, we derive a shelterin score based on shelterin expression. Pathway analysis shows shelterin is positively correlated with E2F targets, while is negatively correlated with p53 pathway. Importantly, shelterin links to tumor immunity and predicts response to PD-1 blockade immune therapy. In-depth miRNA analysis reveals a miRNA–shelterin interaction network, with p53 regulated miRNAs targeting multiple shelterin components. We also identify a significant amount of lncRNAs regulating shelterin expression. In addition, we find shelterin expression could be used to predict patient survival in 24 cancer types. Finally, by mining the connective map database, we discover a number of potential drugs that might target shelterin. In summary, this study provides broad molecular signatures for further functional and therapeutic studies of shelterin, and also represents a systemic approach to characterize key protein complex in cancer.