The ability to manipulate the mouse genome has made the mouse the primary mammalian genetic model organism. It has been possible to model human cancer in the mouse by overexpressing oncogenes or inactivating tumor suppressor genes, and these experiments have provided much of our in vivo understanding of cancer. However, these transgenic approaches do not always completely and accurately model human carcinogenesis. Recent developments in transgenic and knockout approaches have improved the accuracy of modeling somatic cancer in the mouse and analyzing the genomic instability that occurs in murine tumors. It is possible to use retroviral gene delivery, chromosome engineering and inducible transgenes to selectively manipulate the genome in a more precise spatial and temporal pattern. In addition, the development of powerful cytogenetic tools such as spectral karyotyping, fluorescence in situ hybridization and comparative genome hybridization have improved our ability to detect chromosomal rearrangements. Finally, global patterns of gene expression can be determined by microarray analysis to decipher complex gene patterns which occur in cancers. Several of these advances in mouse modeling of human cancer are discussed in this review.