Three experiments support the hypothesis that mechanisms involved in observational conditioning (OC) of fear are similar to those of direct classical conditioning and involve the organism attempting to detect the causal structure of its environment. Experiment 1, a correlational analysis, shows that model monkeys' fear behaviors on snake trials (unconditioned stimulus [US]) were highly correlated with observer monkeys' fear (unconditioned response) while watching the models' fear. In Experiment 2, all observers showed distress while watching the model's fear during Session 1 of OC, but only observers who could see the snake to which the model was reacting continued to show fear during subsequent OC sessions, suggesting that the model's fear is an easily habituable US. In Experiment 3, observers acquired significant fear of snakes after 1 OC session, indicating that the continued fear of those Experiment 2 observers that could see the snake may reflect their own acquired fear of snakes.