A general problem in biology is how to incorporate information about evolutionary history and adaptation into taxonomy. The problem is exemplified in attempts to define our own genus, Homo. Here conventional criteria for allocating fossil species to Homo are reviewed and are found to be either inappropriate or inoperable. We present a revised definition, based on verifiable criteria, for Homo and conclude that two species, Homo habilis and Homo rudolfensis, do not belong in the genus. The earliest taxon to satisfy the criteria is Homo ergaster, or early African Homo erectus, which currently appears in the fossil record at about 1.9 million years ago.