We analysed phylogeography and population genetic variation across the range of the western pond turtle (Emys marmorata) using rapidly evolving mitochondrial and nuclear DNA sequence data. Nuclear DNA sequences from two unlinked introns displayed extremely low levels of variation, but phylogenetic analyses based on mtDNA recovered four well-supported and geographically coherent clades. These included a large Northern clade composed of populations from Washington south to San Luis Obispo County, California, west of the Coast Ranges; a San Joaquin Valley clade from the southern Great Central Valley; a geographically restricted Santa Barbara clade from a limited region in Santa Barbara and Ventura counties; and a Southern clade that occurs south of the Tehachapi Mountains and west of the Transverse Range south to Baja California, Mexico. An analysis of molecular variance (amova) based on regional hydrographic units revealed that populations from the Sacramento Valley north to Washington were virtually invariant, with no evidence of population substructure among northern river drainage basins. In other areas, E. marmorata contains considerable unrecognized variation, particularly in central and southern California and in northern Baja California, Mexico. Our northern clade is congruent with the distribution of the subspecies Emys marmorata marmorata (Washington–central California). However, no clade is congruent with the distribution of the southern subspecies Emys marmorata pallida from central California–Baja. Thus, recognition of the current subspecies split is not warranted, based on the available genetic evidence. Our amova and phylogenetic results, in conjunction with a growing comparative database for other codistributed aquatic taxa, confirm the occurrence of genetic breaks across the Tehachapi Mountains and Transverse Range bounding the southern end of the Great Central Valley, and point to southern California as a rich source of cryptic genetic variation.