Observational learning is an important skill for all children to acquire. Children with autism often do not demonstrate this skill nor do they learn it on their own. The present study, using a multiple baseline across participants, single case, research design, investigated the effects of using a peer-yoked contingency game with four male participants with autism, aged 4-7 years. Each participant was presented with a simple labeling task while his friend was seated beside him. Participants had the same partners throughout the treatment. Once the model response was emitted, the teacher presented the same task to the observing boy. Data were collected on correctly observed and emitted responses during the game. Pre- and post probes and tests were conducted for observational learning, generalized imitation, and learned reinforcement for peers. Results from this study provide support for the use of the peer-yoked contingency game as a method for increasing observational learning in children with autism. All four participants increased their correct responding to specific tasks and increased their demonstration of observational learning in a natural educational setting. Evidence of increased interest in peers was also observed. The present study provides support for the use of the peer-yoked contingency game to teach observational learning.