We propose that similes involve a form of pragmatic reasoning that has been overlooked in the literature: the derivation of a scalar implicature. When someone utters the simile ‘Fred is like a lion’ (or a literal comparison: ‘Wilma is like a nurse’), they normally imply that Fred is not a lion, otherwise they should have used the stronger categorization statement ‘Fred is a lion’. We report 3 experiments showing that children as young as 3 years are able to derive the scalar implicature ‘x is not a y’ when interpreting similes of the form ‘x is like a y’.