Karyotypic abnormalities have been described in more than 10,000 human neoplasms analyzed by means of chromosome banding. These aberrations are of three different kinds: primary abnormalities, which are essential in establishing the tumor; secondary abnormalities, which develop only after the neoplasm is established but which nevertheless may be important in tumor progression; and cytogenetic noise, which is the background level of nonconsequential aberrations. These latter changes are, in contrast to the primary and secondary aberrations, randomly distributed throughout the genome. The primary abnormalities, of which more than 100 have been identified, are strictly correlated with particular neoplastic disorders and even with histopathological subgroups within a given tumor type. To these purely cytogenetic data implicating specific genetic changes in carcinogenesis may now be added the growing evidence of molecular specificity emerging from recombinant DNA studies. It appears that both currently known classes of directly cancer-relevant genes, the dominant oncogenes and the recessive anti-oncogenes, are located at precisely those genomic sites that are visibly involved in neoplasia-associated chromosomal rearrangements. The molecular genetic data thus support the cytogenetic conclusion that the distribution of consistently cancer-associated breakpoints reflects the genomic position of genes that, either directly or through the control function they exert, are essential in the proliferation and differentiation of human cells.