Childhood altruism is a widely explored domain, where a majority of literature claims that children are overtly prosocial beings. It has been demonstrated that young children extend prosocial behaviour towards others irrespective of personal cost. Further, children consistently calculate and utilise social information (e.g., the morality of a character) to guide their prosocial actions. The aim of the current thesis was to determine whether morally contingent actions committed by a protagonist impact children’s degree of help extended when this incurs a marked cost. A final sample of 42 children aged between four- and five-years was presented with two puppets. One puppet behaved morally, immorally or neutrally towards the other puppet. Children were then presented with two boxes containing one and three rewards. They were asked to allocate one reward to the morally contingent puppet and one to themselves. While both options are prosocial, allocating the puppet three items is altruistic and costly. Contrary to the claims of previous findings, children did not act in an altruistic manner but instead consistently chose the larger reward for their own keeping. Further, prosocial behaviours were not affected by the morality of the character. These results support evolutionary and social accounts of altruism and suggest that, when given multiple prosocial options, children will reliably choose to monopolise rewards. The findings of the current study call into question previous claims that children are strongly, inherently altruistic.