An adapted alternating treatments design was used to compare the efficacy of echoic and textual prompts to teach three students with autism (ages 8–15) to ask questions related to two pre-selected topics of conversation. Participants were first required to answer questions related to the topics to determine whether accurate responses were within their repertoire. This was followed by a transfer of stimulus control procedure to teach participants to ask relevant questions to the experimenter on the same topics. Probes with a novel conversation topic were conducted in the natural environment with a peer, and follow-up probes were conducted after training. Results indicate questions taught using an echoic prompt were acquired in fewer trials to criterion than questions taught using a textual prompt for two of the three participants. Limitations and implications for future research will be discussed.