The modeling effect on eating means that the more models eat the more participants eat. Herman, Polivy, & Roth (2003) proposed that participants make the amount of food they consume conform to the consumption of others in order to avoid being seen by others as eating excessively. In this study, in order to create a situation in which participants believe no one can know how much they eat, we did not use the usual model. Instead, feigned leftover food was shown to participants before the tasting test. This leftover food (in amounts large or small) was expected to give participants information on how much other participants had eaten. In one condition, participants were misled to believe that the experimenter could not find out how much food the participants had consumed; in another condition, they were not misled. In the former condition, regardless of how much others eat, participants should eat as much as they like, believing that no one can learn of the amount, they consume. Contrary to the prediction, the modeling effect arose in both conditions. These results indicate that the modeling effect cannot be explained entirely by self-presentational concern regarding others.