Cancer-related investigations have long been in the limelight of biomedical research. Years of effort from scientists and doctors worldwide have generated large amounts of data at the genome, transcriptome, proteome and even metabolome level, and DNA and RNA cancer signature databases have been established. Here we present a database of differentially expressed proteins in human cancers (dbDEPC), with the goal of collecting curated cancer proteomics data, providing a resource for information on protein-level expression changes, and exploring protein profile differences among different cancers. dbDEPC currently contains 1803 proteins differentially expressed in 15 cancers, curated from 65 mass spectrometry (MS) experiments in peer-reviewed publications. In addition to MS experiments, low-throughput experiment data from the same literatures and cancer-associated genes from external databases were also integrated to provide some validation information. Furthermore, dbDEPC associates differential proteins with important structural variations in the human genome, such as copy number variations or single nucleotide polymorphisms, which might be helpful for explaining changes in protein expression at the DNA level. Data in dbDEPC can be queried by protein identifier, description or sequence; the retrieved protein entry provides the differential expression pattern seen in cancers, along with detailed annotations. dbDEPC is expected to be a reference database for cancer signatures at the protein level. This database is provided at http://dbdepc.biosino.org/index/.