The current study investigated the effect of observational learning during equivalence based instruction (EBI). Two boys (Tim and Nate) ages 11 and 12 with Autism Spectrum Disorder participated in the study. Participants received small-group EBI training with an embedded observational learning component twice weekly for six weeks. Both participants were given a trained and observation set containing three classes (Class A, Class B and Class C) consisting of four class members. Participants served as both learners and observers during each training session. Each participant was trained on match-to-sample tasks with relations A-B and B-C and tested for class formation across the trained and observation set. Results showed that Tim was able to derive the untrained A-C and C-A relation at 100% correct on both the trained and observation set of stimuli. After the initial training, Nate averaged at 40% and 55% on the trained and observation set of stimuli, indicating that he was unable to derive the untrained relations. Two remedial training sessions were conducted, where Nate was re-exposed to the A-B and B-C training. After the remedial training, Nate averaged at 85% and 67.5%, indicating strong class formation on the trained set of stimuli, and moderate class formation on the observation set. The current study demonstrated the utility of observational learning during EBI. Limitations and implications for clinical practices are discussed.