This study showed that a means-end relationship between two snacks exerts a negative influence on preference for the means snack in the contingency. Two snacks ranked of approximately equal medium appeal were individually chosen from an array by each of 86 children (ages 4 years, 4 months to 7 years, 2 months). Children were then assigned to a means-end condition (where one snack was eaten as a means of gaining the other), a temporal order control group (where one snack was presented and eaten before the other), or a mere exposure control group (where the children chose the order in which they ate the two snacks). Post-treatment rankings of the snacks showed that children in the means-end group came to devalue the means snack relative to the reward snack, despite the demonstrably equal prior appeal of the two snacks. Children in the temporal order and mere exposure control groups did not come to prefer the two snacks preferentially. Implications of the finding that adults can affect children's preferences by relatively minor variations in the manner of presentation are discussed.