Since cancers have individual clonal karyotypes, are immortal and evolve from normal cells treated by carcinogens only after exceedingly long latencies of many months to decades-we deduce that carcinogenesis may be a form of speciation. This theory proposes that carcinogens initiate carcinogenesis by causing aneuploidy, i.e., losses or gains of chromosomes. Aneuploidy destabilizes the karyotype, because it unbalances thousands of collaborating genes including those that synthesize, segregate and repair chromosomes. Driven by this inherent instability aneuploid cells evolve ever-more random karyotypes automatically. Most of these perish, but a very small minority acquires reproductive autonomy-the primary characteristic of cancer cells and species. Selection for autonomy stabilizes new cancer species against the inherent instability of aneuploidy within specific margins of variation. The speciation theory explains five common characteristics of cancers: (1) species-specific autonomy; (2) karyotypic and phenotypic individuality; (3) flexibility by karyotypic variations within stable margins of autonomy; (4) immortality by replacing defective karyotypes from constitutive pools of competent variants or subspecies generated by this flexibility; and (5) long neoplastic latencies by the low probability that random karyotypic alterations generate new autonomous species. Moreover, the theory explains phylogenetic relations between cancers of the same tissue, because carcinogenesis is restricted by tissue-specific transcriptomes. The theory also solves paradoxes of other cancer theories. For example, "aneuploidy" of cancers is now said to be a "paradox" or "cancer's fatal flaw," because aneuploidy impairs normal growth and development. But if the "aneuploidies" of cancers are in effect the karyotypes of new species, this paradox is solved.