The present study examined social inference in rats about the availability of food based on the behavior of conspecifics and the types of food sites. First, rats were trained to learn two types of food sites: replenished and depleted sites. On the first trial of a daily training, rats were allowed to eat at four food sites in an experimental field. On the second, third, and fourth trials, two food sites were replenished with foods (replenished sites), but the other two sites were not (depleted sites). All rats learned the type of the food sites and came to visit the replenished sites earlier than the depleted sites on the second and later trials. In the test conducted after the training, another demonstrator rat was shown to the subjects as information in a social inference situation. That is, the demonstrator rat first visiting one of two types of sites: either the replenished site or the depleted one. Subsequently, a subject rat was given a choice between the two sites of the same type: the site previously visited by the demonstrator and another site that was not visited. Subject rats did not show any sign of social inference; that is, they were not able to infer the existence of food based on an integration of knowledge about types of food sites as well as conspecifics' behavior. They showed a tendency to follow the demonstrator regardless of the type of the food sites. Although these results suggest absence of social inference ability for combining information about behavior of other conspecifics and type of food sites, a discussion is entertained that rats' conformity to others might override personal knowledge.