The purpose of this study was to evaluate the effects of different speech-generating device displays and vocabulary organizations on the acquisition of multi-step requesting responses in children with autism spectrum disorder (ASD). Four young children with ASD were taught to use an iPad® application to make requests using both a taxonomically organized grid display and a schematically organized visual scene display or hybrid. The conditions were compared using a multi-element design. Time delay and least-to-most prompting were used to teach responses in both conditions. Three participants met mastery criterion for acquiring requests with the schematic display but did not meet criterion requesting with the taxonomic display. A fourth participant learned to make requests with both displays but showed generalization only with the schematic display. Error analyses indicated there were different types of errors made across conditions, which may suggest the need to tailor intervention methods to specific displays. Implications for the design of AAC [augmentative and alternative communication] displays, assessment, and interventions are discussed.