Theories proposing that human language evolved as gesture rather than speech have garnered increasing interest in the last decade, fueled by studies showing that children and great apes use gesture in relatively sophisticated ways. Proponents of “gesture first” theories further argue that gesture could have provided a communicative advantage to human ancestors, even in the absence of a shared communicative code, because it has the potential to convey meaning iconically. Such manual iconicity (often referred to as pantomime) could theoretically be invented on the spot and could be understood by others without any shared linguistic system. This is a compelling story, but while gesture was a likely component in the communication of early human ancestors, it is unlikely that pantomime played a central role. Although iconic signs seem in many ways to be the most intuitive of Peirce’s trichotomy, understanding and producing iconic signs requires complex mental representation and relatively sophisticated analogic reasoning. Children do not begin to use iconic gestures until well after they have mastered both conventional and indexical gestures and have begun acquiring speech. Apes show great facility with gesture in general, but there is little evidence that they use or understand iconic signs. I will present data on the emergence of iconic communication and conventional linguistic structures in hearing children learning English and a profoundly deaf child creating his own manual language (i.e. “homesign”). I will compare these developmental trajectories to the gestural communication of great apes, arguing that human children (but not apes) come to understand iconicity at around age 3. While the ability to manipulate abstract, conventional symbols is unquestionably one of the key capacities of our species, the ability to understand the abstract mapping between an iconic sign and a referent may require a similarly sophisticated mind.