Children with autism spectrum disorder (ASD) have been shown to exhibit fewer instances of social referencing compared to their typically developing peers. The current study evaluated the effectiveness of multiple-exemplar training, prompting, and social reinforcement to teach social referencing. We used a single-blind randomized control trial with a multiple-baseline design nested within the experimental group to evaluate treatment effects. Twenty-five children with ASD participated. Participants in the treatment group showed significantly higher social referencing scores. However, there were no statistically significant differences between the groups in the posttraining initiating and responding to joint attention scores. The clinical implications of our findings are discussed.