The inconsistencies which exist when the traditional tooth-to-bony-landmark/tooth-to-occlusal-relationship criteria of identification of teeth are maintained are discussed. It is pointed out how these (e.g., "the canine is the tooth behind the premaxillary-maxillary suture") can be falsified. It is also suggested that some mammals, including Tarsius and Homo sapiens, develop homologies of three sets of "teeth," and that the "adult" antemolar dentition of a mammal may be composed of retained deciduous teeth as well as permanent teeth. Following a revision of dental homologies in most primates, an approach to reevaluating dental homologies is proposed, and a model of tooth "loss" presented.