We present an approach making use of technology established in the context of the genome project to describe a pancreatic cancer-specific expression profile and to identify new potential disease genes or disease-associated-genes. By use of gridded arrays of pancreatic cancer cDNA libraries and differential hybridizations we show that 4% the gridded cDNA library clones contain sequences preferentially expressed in pancreatic cancer. EST-sequencing of 369 distinct (408 total), differentially expressed sequences identified novel genes (32.5%) or homologs to EST-sequences with unknown function (26.3%). Homologies to known genes allow to determine a pancreatic cancer-specific expression profile, which provides for the first time evidence for complex primary and secondary alterations of gene expression responsible for the development of the phenotype of pancreatic cancer cells. In addition this has led to the identification of novel differentially expressed genes, which represent potential oncogenes or disease-associated markers and may be helpful for the development of therapeutic or diagnostic modalities.