Gossypium hirsutum L., the most important species of cultivated cotton, is a diverse species whose variation patterns reflect both natural phenomena and human selection and trade. Analysis of molecular marker variation, in the context of historical data, assisted in resolving the confounding affects of human domestication that heretofore have obscured the taxonomic and phylogenetic relationship between G. lanceolatum and G. hirsutum, natural patterns of interspecific gene flow between G. barbadense and G. hirsutum, and the geographical origin of domesticated G. hirsutum. In addition to uncertainty about its taxonomic circumscription, Gossypium lanceolatum represents the focal point of an hypothesis that tetraploid Gossypium have a polyphyletic origin. Reevaluation of this hypothesis using historical and molecular markers demonstrates that G. lanceolatum is genetically embedded within and cladistically indistinguishable from G. hirsutum. Thus, G. lanceolatum is more properly recognized as G. hirsutum race 'palmeri' and both 'G. hirsutum sensu lato and tetraploid Gossypium are inferred to be monophyletic. Analysis of diagnostic cpDNA restriction site data and nuclear markers in sympatric and allopatric populations of G. hirsutum and G. barbadense lead to several conclusions: (1) introgression between G. hirsutum and G. barbadense is bidirectional for both nuclear and cytoplasmic genes; (2) patterns of introgression between the two species are not symmetrical; (3) nuclear introgression is geographically more widespread and more frequently detected than cytoplasmic introgression. Complex genetic relationships among G. hirsutum populations obscure the geographical origin of domesticated G. hirsutum. Analyses of allelic variation at 205 RFLP loci implicate the Yucatan peninsula as the site for the earliest stages of domestication. Further development and subsequent dispersal of Yucatan cultigens in southern Mexico and Guatemala may be the cause of the southern Mexico/Guatemala center of diversity traditionally interpreted as the origin of domesticated G. hirsutum. The gene pool of modern Upland cultivars derives from Mexican highland populations that in turn trace their origins to southern Mexico and Guatemala. Despite the wide diversity of introductions involved in the development of the Upland cotton, the genetic base is extremely narrow. This suggests that the more recent Mexican highland gene pool supplanted much of the early germplasm.