This study examined the use of shaping with a percentile schedule to increase the duration of the interaction following a bid for joint attention in children with autism. Specifically, the therapist initiated a bid for joint attention and reinforced longer successive approximations in seconds of sustained interaction with the therapist and activity. A percentile schedule ranked the most recent 10 observations and reinforcement was provided if the current observation equaled the sixth ranking. Most-to-least prompting was used if the child failed to meet the calculated criterion. Shaping with a percentile schedule of reinforcement was effective at increasing the duration of sustained interaction following a bid for joint attention, for all participants from an average baseline duration of 13 s to an average intervention duration of 215 s.