We used an adapted alternating treatments design to compare the effects of traditional and embedded discrete trial teaching (DTT) with adults with autism. Traditional DTT consisted of the instructor presenting a discriminative stimulus to start each trial (“Point to___”), implementing a prompt (i.e., manual guidance), and providing reinforcement (i.e., praise and token) for correct responding. Embedded DTT consisted of presenting this same trial structure within an ongoing activity (e.g., playing bean bag toss). The percentage of correct responses improved with the introduction of both types of DTT; however, for both participants, embedded DTT showed faster acquisition (i.e., fewer trials to criterion) but longer session durations than traditional DTT. Participants showed more positive affect during embedded DTT and chose it more often relative to traditional DTT. These results replicated and extended the results of Geiger et al. (Behav Anal Pract 5(2):49–59, 2012. 10.1007/bf03391823) and have implications for considering participant assent and interest in ABA-based instruction.