On the basis of selected osteological markers, Patterson (1936) identified two pathways (here designated A and B) along which the internal carotid artery in notoungulates was said to travel in order to enter the skull and supply the brain. The existence of these pathways, at least as Patterson defined them, has been disputed but no substantive alternatives have been proposed. Using comparative embryological and morphological evidence based on conditions in extant mammals, we find that pathway A (intratympanic) is not supported. Pathway B (enclosed extratympanic) is supportable, but as far as is now known applies only to a small number of notoungulate taxa. On the basis of new evidence, we propose another route, pathway C, briefly mentioned by Scott (1912) but subsequently ignored, that may apply to the majority of notoungulates. In pathway C (unenclosed extratympanic) the internal carotid passed directly into the endocranium via a naturally unossified area of the basicranium, the piriform fenestra, rather than coursing through or alongside the middle ear in a canal. The absence of a separate, bony carotid foramen on the basicranium's ventral surface may explain the hesitancy of previous workers to consider this routing. There is no evidence that the function of the internal carotid artery was supplanted by another vessel (e.g., external carotid artery) in any notoungulate. Conditions in other major clades of South American native ungulates are poorly investigated, but some clearly differed from notoungulates in carotid patterning, pointing to the existence of substantial intertaxon disparities.