The new giant turtle species Gopherus donlaloi is described from a partial skeleton with skull collected in the Rancholabrean deposits in northeastern México. The skull resembles the skull of extant turtles G. polyphemus and G. flavomarginatus but the new species shows unique skull and shell features. Shell meristic variation of the new species, G. berlandieri, and G. laticuneus is discussed denoting problems in the assessment of species based on extremely variable shell characters. Lack of diagnostic features in shells of G. edae and G. hexagonatus suggests their status as nomina vana. A cladistic strict consensus tree suggests that Gopherus is a monophyletic group where G. mohavetus falls within the outgroup, questioning its status as a member of Gopherus. Oligocene G. laticuneus is sister to all Gopherus, after which Recent G. berlandieri and G. agassizii branched out paraphyletically. Gopherus sensu stricto is monophyletic but the relationships among its taxa are unknown; these include the Miocene G. brevisternus, G. pansus, and G. vagus, the Plio-Pleistocene G. canyonensis and G. donlaloi, and the Recent G. polyphemus and G. flavomarginatus. A second analysis excluding most incomplete taxa retains the polytomy of G. berlandieri, G. agassizii, and Gopherus sensu stricto, but resolves the relationships within Gopherus sensu stricto. G. brevisternus is sister to the rest of the clade, followed by G. flavomarginatus, after which there is a polytomy formed by G. canyonensis, G. donlaloi and G. polyphemus. Bootstrap and branch-support analyses indicate that the clades within Gopherus sensu stricto are well supported. Reanalysis of biogeographic relationships based on the phylogeny suggests that the origin of Gopherus sensu stricto can be traced to the Miocene on the Central Plains, later extending southward from eastern Arizona to Florida and from northern Texas to Aguascalientes during the Plio-Pleistocene. The extinction of giant gopher turtles in Texas and eastern Mexico associated with the invasion of their distribution area by G. berlandieri is the best hypothesis to explain the recent disjunct distribution of G. polyphemus and G. flavomarginatus.