This study examined a behavioral intervention package to promote the use of target vocalizations alongside speech-generating device (SGD) mands. Six minimally verbal children with autism spectrum disorder participated, including three with no prior SGD experience. During baseline, SGD responses resulted in access to a preferred item and there was no reinforcement for vocalizations. In Phase I of intervention, responses that included target vocalizations were differentially reinforced with a highly preferred target item and, following a delay, responses without vocalizations produced an easy distractor trial and access to a lesser-preferred item. Three participants increased vocalizations with these procedures alone. For two of these participants, closer approximations or full words were modeled and differentially reinforced during a secondary intervention phase, resulting in an increase in closer matches for one of them. The third did not require this intervention, but a second target was introduced successfully. Although the remaining three participants responded minimally to Phase I, vocalizations increased to high levels for two of these three participants after a vocal model was added and faded during Phase II. Independent SGD maintained throughout all phases of the study for all participants, and participants generalized the use of vocalization responses when the SGD was not present.