Augmentative and alternative communication systems are widely recommended for nonvocal developmentally disabled individuals, with selection-based systems becoming increasingly popular. However, theoretical and experimental evidence suggests that topography-based communication systems are easier to learn. This paper discusses research relevant to the ease of acquisition of topography-based and selection-based systems. Additionally, current practices for choosing and designing communication systems are reviewed in order to investigate the extent to which links have been made with available theoretical and experimental knowledge. A stimulus equivalence model is proposed as a clearer direction for practitioners to follow when planning a communication training program. Suggestions for future research are also offered.